If this were a 150-point exam, scoring 100 is easy and pushing to 120 is hard; every extra point beyond that is a battle with your own limits. DSD is those last few points — it genuinely exists and can open a gap, but its cost, its boundaries, and whether it's worth it are far more complex than marketing copy suggests.
Companion project: DpdoEngine v6.60
Word count target: 50,000+ words · 130 Q&As

Preface

Since Sony and Philips jointly launched DSD (Direct Stream Digital) in 1996, the debate around it has never stopped. On Chinese-language internet, DSD is deified as "the ultimate digital format with analog flavor"; on technical forums, it's dismissed as "a marketing gimmick." Both extremes are too far out; the middle ground — rational, quantifiable, bounded assessment — is unoccupied.

This article is not meant to deny DSD, nor to praise PCM. The goal is to help readers build a rational judgment framework amid the fog of information. The goal is to sort through the 130 most common DSD misconceptions and give verifiable facts for each, helping readers build their own judgment framework.

The article draws heavily on DpdoEngine's measured data — as one of the few complete toolchains on consumer hardware supporting both PCM→DSD (main mode) and DSD→PCM (--reverse mode, based on the Pdm2Pcm engine), we accumulated rich comparison data during development. DpdoEngine's digital filters use AVX-512/AVX2/NEON multi-platform SIMD acceleration, supporting up to 131072-tap FIR filtering (--hb mode) and Kakeya polynomial coefficient compression — the default 65536-tap FIR compresses to a 128th-order polynomial (1/512); --hb mode compresses 131072 taps to a 64th-order polynomial (1/2048), achieving strict quantitative control between DSD-domain noise shaping and PCM-domain precision. This data is not marketing copy; it is reproducible.

About this article's positioning: Written by the DpdoEngine development team. DpdoEngine is one of the few complete toolchains on consumer hardware supporting bidirectional PCM→DSD and DSD→PCM conversion. Writing a consumer-facing format explainer may seem to contradict product interest, but it does not — DpdoEngine's core philosophy is: effective use of DSD must be based on understanding its real technical boundaries, not as a substitute for marketing narrative. We don't believe DSD is "magic"; it's a product of engineering trade-offs. That's precisely why we ensure every parameter on the conversion path is transparent and verifiable. This article targets audio enthusiasts and professionals who want to rationally evaluate DSD's value — fully aligned with DpdoEngine's product positioning.

Chapter 1: Misconceptions About Technical Nature (15 items)

Q1: Is DSD a brand-new "non-PCM" technology?

Misconception: From sampling principles to encoding, DSD is completely different from PCM — a third path in digital audio.

Truth: DSD uses pulse-density modulation (PDM). It belongs to the same sampling-encoding framework as PCM, but differs fundamentally in encoding: PCM quantizes the signal's amplitude at each sampling instant, while DSD encodes amplitude *changes* through the density of 1-bit pulses. Calling DSD "1-bit PCM" is a simplification that's not quite accurate; its characteristics and limits also differ sharply from multi-bit PCM.

From an information-entropy view, DSD64 stereo's raw bitrate is 5.6448 Mbps, while CD-quality 16-bit/44.1kHz stereo is 1.4112 Mbps. DSD64's bitrate is 4× CD. Given that it must carry a full audio band (20Hz-20kHz), its information density is lower than multi-bit PCM, but higher than general intuition at equal bitrate.

Why this matters: Many believe DSD's "non-PCM" status means it's some more advanced encoding system — the foundation of the entire DSD myth. Once you understand DSD is PDM (pulse-density modulation) — same digital sampling framework as PCM but a different encoding — one amplitude-modulated, one density-modulated — all subsequent technical discussion gains the correct coordinate system.

Q2: Is DSD's "1-bit" design a symbol of advancement and precision?

Misconception: 1-bit means the simplest, most direct conversion and the purest signal path, hence the highest precision.

Truth: 1-bit is actually one of the coarsest quantization methods. A single bit can only represent "above threshold" or "below threshold." At 1 bit, a sample period's theoretical dynamic range is just 20·log₁₀(2) ≈ 6.02 dB — three orders of magnitude below CD's 96dB theoretical dynamic range.

Why does DSD work at all? Because it uses extreme oversampling rates (64× = 2.8224MHz) plus noise shaping to drive quantization noise out of the audible band into the ultrasonic range, "squeezing out" about 120dB of effective dynamic range within the 20Hz-20kHz audible region. This process itself is complex and nonlinear — not "the simplest path."

Q3: Does a high sample rate (e.g., 2.8MHz) equal high sound quality?

Misconception: DSD64's 2.8224MHz rate far exceeds CD's 44.1kHz, so DSD64 contains far more information than CD.

Truth: Sample rate is just one dimension of digitization. The other key dimension is quantization depth (bit depth). DSD64 is 2.8MHz but each sample has only 1 bit. Quantization noise is uniformly distributed across the entire Nyquist band (0-1.4112MHz); without noise shaping, the in-band noise floor would be absurdly high.

Specifically:

DSD64's raw quantization noise power in the same band is about 63dB higher than PCM 16-bit
Noise shaping concentrates energy above 20kHz, bringing in-band SNR to ~120dB
But the cost: 20kHz-1.4MHz is filled with extremely high-intensity noise

Calling DSD64 a "2.8MHz high-sample-rate format" is like calling a sensor with huge noise a "high-resolution sensor" — the sample rate is genuinely high, but the effective information density is not necessarily higher than multi-bit systems at far lower rates.

Q4: Does DSD have no quantization error?

Misconception: Since DSD is 1-bit with only 0s and 1s, the concept of quantization error doesn't exist.

Truth: Every digital system has quantization error. A 1-bit system's quantization error is *more severe* than multi-bit — each sample can only output one of two opposite states, and the RMS gap from the true analog signal (quantization error) is enormous over short intervals.

DSD cleverly uses noise shaping to change the quantization error's spectral distribution, concentrating energy out-of-band — but it doesn't eliminate the error itself. The residual in-band quantization noise power still defines DSD's noise floor — around -120dBFS for DSD64, its limit.

Q5: Does DSD have no linearity error?

Misconception: 1-bit DSD DAC conversion is "naturally linear" because it only switches between two voltages.

Truth: DSD's D/A conversion (turning the 1-bit stream into analog voltage) is indeed less susceptible to resistor-precision issues than multi-bit R-2R ladders in principle — but that doesn't mean no linearity error. DSD's linearity errors come from:

Pulse asymmetry: an ideal square wave's rising and falling edges should be symmetric; real circuits differ, causing delay jitter and correlated distortion
Modulation noise: nonlinearities in noise-shaping algorithms produce distortion components correlated with the input signal
Switching jitter: the 1-bit stream switches at extremely high rates (2.8MHz+), so clock jitter significantly affects sound quality

These nonlinear factors are long-studied topics in DSD DAC design — not "naturally linear."

Q6: Is DSD's dynamic range inherently huge?

Misconception: DSD's nominal 120dB dynamic range exceeds CD's 96dB, so DSD records more extreme soft-loud contrasts.

Truth: DSD's 120dB dynamic range isn't inherent — it's "borrowed" through noise shaping. Without noise shaping, relying only on oversampling gain (OSR = 64, contributing ~18dB), DSD64's dynamic range within 20Hz-20kHz is about 24dB. Noise shaping isn't "icing on the cake" — it's the core competency; without high-order noise shaping, DSD can't compete with PCM 16-bit. You need 5th-7th-order noise shapers to push in-band noise down to -120dBFS.

This is fundamentally different from PCM: PCM 16-bit's 96dB dynamic range is inherent — no extra processing needed; each 16-bit sample's resolution alone determines 96dB. PCM 24-bit's inherent dynamic range is 144dB, far exceeding DSD64. Noise shaping in PCM is a nice-to-have (improving quantization noise spectral distribution); in DSD it's a necessity — without it, DSD is unusable.

Q7: Does DSD's effective dynamic range far exceed PCM?

Misconception: DSD64's nominal 120dB vs CD's 96dB means DSD is better.

Truth: This comparison is one-sided — it pits DSD64's highest spec against PCM's lowest.

Fair comparison:

Format	Theoretical dynamic range (20Hz-20kHz)
CD (PCM 16/44.1)	96 dB
DSD64	≈120 dB (after noise shaping)
PCM 24/96	144 dB (no processing needed)
PCM 24/192	144 dB (no processing needed)
DSD128	≈130 dB+ (significantly above DSD64 with 7th-order shaping)

DSD64's audible-band dynamic range indeed beats CD, but sits far below any PCM 24-bit spec. Moreover, DSD's dynamic-range ceiling is limited by the noise shaper's dynamic precision; effective dynamic range drops sharply at high frequencies.

Q8: Is DSD the ultimate "lossless" format?

Misconception: "DSD is the digital master" — it's lossless, unlike lossy MP3 compression.

Truth: DSD isn't lossy in the "lossy compression" sense, but it *is* lossy in digitization — every A/D conversion loses something, DSD included. More critically, DSD performs extremely aggressive noise shaping to achieve usable SNR within the 1-bit framework — itself a form of signal-processing loss.

"Lossless" should mean minimal difference between the digital master and the original analog signal. DSD64's -120dB noise floor is 24dB worse than 24-bit PCM's -144dB (about 15× more noise energy). By "digital master" standards, DSD64 is even behind an ordinary 24-bit PCM file on dynamic range.

Q9: Does DSD's noise shaping eliminate noise?

Misconception: Since noise is "shaped" out of the audible band, inaudible equals absent.

Truth: Noise shaping doesn't eliminate noise; it redistributes the noise spectrum. Total noise energy is unchanged (per Bennett's quantization-noise theory, noise power is constant) — just pushed from low to high frequencies.

The most direct evidence: a DSD DAC's output needs a steep analog low-pass filter (typically cutoff 50-80kHz) to remove out-of-band noise. Without it, DSD's high-frequency noise could burn tweeters or cause amplifier self-oscillation. The noise isn't "gone" — it's "relocated" to a place that needs a dedicated filter to contain.

Q10: Is DSD's out-of-band noise irrelevant?

Misconception: Since humans can't hear above 20kHz, DSD's ultrasonic noise has no effect.

Truth: Ears can't hear ultrasound, but ultrasonic noise can:

Cause intermodulation distortion (IMD): in amplifier/DAC nonlinear regions, ultrasonic signals mix with audible-band signals, producing new audible distortion
Saturate amplifiers: high-power ultrasonic noise consumes dynamic headroom, causing premature clipping
Affect drivers: tweeter mechanical resonances can fall in the ultrasonic range, causing after-ringing
Trigger speaker protection: some active speakers' DSP detects ultrasonic energy and engages protection

Under DpdoEngine v6.60 default config (taps=65536, order=7), DSD64 output's residual noise total energy in 20kHz-100kHz is about 30dB above PCM 24/96's quantization noise floor. This varies with modulator design and output config. This means DSD's downstream gear needs sufficient headroom for this noise, or quality actually degrades.

Q11: Does the name "Direct Stream" mean a simpler signal path?

Misconception: "Direct Stream Digital" means direct digital streaming from recording to playback — shortest path, least information loss.

Truth: The "Direct Stream" name is largely marketing rhetoric. In real systems, DSD's signal path is often *more* complex than PCM:

Recording end: most DSD recordings use ADI (Audio Data Interchange) modules — internally multi-bit ΔΣ modulators outputting multi-bit streams, then decimated and re-modulated into 1-bit streams — already two conversions
Editing end: DSD can't be edited directly; must convert to PCM or multi-bit DSD (e.g., DXD), edit, then convert back to 1-bit DSD — two more conversions
Playback end: many DACs convert DSD to multi-bit PCM internally before D/A — not true "direct pass"

Hypex founder Bruno Putzeys once noted: "DSD's 'direct' is a marketing term; technically it may be one of the most circuitous digital audio paths today."

Q12: Is DSD a major revolution in audio technology?

Misconception: DSD and PCM are two independent parallel technology lines representing a breakthrough in digital audio.

Truth: DSD is essentially a compromise — extreme oversampling plus noise shaping forced by the need for high fidelity within the 1-bit framework. This had some rationale in the mid-1990s (when multi-bit DAC linearity and cost were hard to balance), but it's hardly a "revolutionary breakthrough."

From Philips's 1987 introduction of 1-bit D/A conversion to the 1996 SACD/DSD standard, DSD's core idea (1-bit + noise shaping + ΔΣ modulation) has always been a branch application of ΔΣ modulator technology, not an independent revolution. The theoretical foundation of ΔΣ modulation dates to 1962 (Inose, Yasuda and Murakami's paper).

Q13: Is DSD the most-used mastering format in studios?

Misconception: Professional studios prefer DSD as the mastering format because it sounds best.

Truth: The studio mastering industry standard is PCM 24-bit/96kHz (or 24/192kHz). The AES (Audio Engineering Society) recommended practice (AES5-2008) is PCM-based. Mainstream DAWs (Pro Tools, Logic Pro, Cubase, Ableton Live) all run PCM engines.

The closest thing to "native DSD" recording format is DXD (Digital eXtreme Definition) — 24-bit/352.8kHz PCM. The name contains "DSD," but it's high-spec PCM. DXD exists precisely because engineers found DSD can't be edited directly and post-production must happen in the PCM domain.

Albums genuinely recorded all-native DSD (including editing and mixing) are estimated at under 1% of global releases.

Q14: Is DSD more "high-fidelity" than high-spec PCM?

Misconception: On high-end systems, DSD "clearly outperforms" 24/96 or 24/192 PCM files.

Truth: By quantifiable fidelity metrics:

THD+N: excellent DACs in PCM 24/192 mode hit <-110dB; same unit's DSD64 mode is about -95 to -105dB. This difference mostly reflects how well the DAC's analog circuitry is optimized per mode, not a fundamental format gap
Dynamic range: PCM 24/192 144dB (inherent) vs DSD64 120dB (after noise shaping)
Frequency response flatness: PCM is usually within ±0.1dB across 20Hz-20kHz; DSD may show rolloff or phase shift near 20kHz depending on low-pass design
Intermodulation distortion: DSD's out-of-band noise produces measurable IMD increments in amplifier nonlinear regions

These measurements show DSD has limits on pure metrics — but objective measurement doesn't directly equal listening value. The question isn't "can DSD see" but "are these measurable differences audible under controlled conditions" — the two must not be conflated.

Q15: Was 1-bit conversion the most advanced technical choice of its time?

Misconception: Sony and Philips chose 1-bit DSD for SACD because it was the most advanced technical direction.

Truth: When the SACD standard was set in 1995, Sony's primary motive was not sound quality but content protection. CD's weak copy protection had been fully cracked by the 1990s; Sony needed a new physical carrier to regain control of distribution channels. SACD's encryption system (Pit Signal Processing, PSP) and physical anti-copy layers (including hidden watermarks) were the core requirements.

Technically, Philips held extensive patents in 1-bit DAC (BitStream); Sony had advantages in DVD physical format. Their collaboration bound both patent pools into the next high-res audio standard — a typical commercial-technological alliance. 1-bit conversion wasn't "the most advanced choice" but "the optimal commercial solution covering both patent pools while satisfying anti-piracy."

*EQ and dynamics processing in mastering**

The subwoofer/driver performance, cabinet design and amplifier drive capability of the playback system
The acoustic treatment of the listening room

The digital format itself (DSD or PCM) records 20-200Hz lows with precision far beyond any playback system's physical limits. DSD64's noise floor at 20Hz is about -120dBFS; PCM 24-bit is -144dBFS — the practical meaning is that both formats record signals far below a listening room's background noise (typically 30-40dBA). Any "format-caused low-frequency difference" sits below the room's noise floor.

Q26: Is DSD's vocal reproduction more "poisonous" and intimate?

Misconception: Listening to audiophile vocals like Teresa Teng or Tsai Chin on DSD sounds more "toxic," more "emotional."

Truth: "Toxic" is a highly subjective term in vocal reproduction, possibly pointing to one of three measurable sound characteristics:

Moderate mid-band (500Hz-3kHz) emphasis (an EQ effect)
Slightly increased even-order harmonic distortion ("warming")
Compressor/limiter use in mastering

If a perceived difference genuinely exists (and isn't blind), the more likely causes:

The DSD version and CD version come from different masters — different mastering engineers and chains
DSD releases often undergo special mastering to highlight "high-res" advantages
DSD upsampling changes the signal spectrum — some upsampling algorithms add extra harmonics

DpdoEngine's round-trip test shows: after a PCM file goes through DSD64 and back to PCM, the FFT difference from the original is below -110dB — the vocal spectrum structure undergoes no measurable change. Any "more toxic" perception is most likely not caused by the DSD format itself.

Q27: Is DSD's instrument separation higher?

Misconception: During DSD playback, instruments separate better and every part is clearly distinguishable.

Truth: With good mic placement and sensible mixing, any format (even MP3 320kbps) guarantees good separation. The format's main effect is loss rather than gain — highly compressed lossy formats drown parts; DSD and PCM as lossless formats both preserve the full image.

In DSD-vs-PCM ABX comparisons requiring subjects to count instruments, studies reached the same conclusion: no significant difference in part resolvability between formats.

Q28: Can DSD let you hear the studio as it was?

Misconception: With DSD, listeners hear exactly what the engineer heard in the control room.

Truth: This has nothing to do with DSD and everything to do with monitoring-chain consistency. Engineers monitor via:

Room-calibrated monitor speakers
Specific DAC/amplifier combinations
A calibrated listening position

Home listeners use completely different systems (different DAC, speakers, room). No format can eliminate playback-chain differences. DSD's claimed "original appearance" is marketing rhetoric — technically impossible.

More critically: most "DSD records" passed through PCM at least once during production (editing, mixing) — the "original" was never DSD to begin with.

Q29: Can upsampling to DSD "create" quality out of nothing?

Misconception: Upsampling PCM to DSD with HQPlayer or similar tools makes ordinary recordings sound better.

Truth: Upsampling can change time-domain interpolation but cannot add information. The Nyquist-Shannon sampling theorem already bounds it: if the original signal was correctly digitized at its sample rate, upsampling only changes the representation of data points — it cannot recover content beyond what was captured.

What upsampling to DSD can do:

Use superior interpolation filters (e.g., sinc) to improve reconstruction precision in the digital domain
Change quantization noise spectral distribution via noise shaping
Play ordinary 16/44.1 PCM as DSD, exploiting the DAC's different analog output stage in DSD mode

What it cannot do:

Introduce high-frequency information not in the original recording
Remove the original's quantization noise
Fix distortion from poor A/D conversion

A special note: some upsampling algorithms deliberately add "seasoning" (e.g., harmonic exciter effects) that users mistake for "quality improvement." When upsampling with DpdoEngine, we default to precise FIR upsampling and offer DSEE (spectral extension) and DSRE (downsampling-residue extension) as optional enhancements — the former adds no extra processing; the latter lets users consciously add harmonic coloration. Know what you're listening to.

Q30: Can't hear DSD vs CD because the gear isn't good enough?

Misconception: You can only hear DSD's benefits on systems above ¥100,000; cheap systems lack resolving power.

Truth: This is a classic marketing-trap ladder:

You can't hear the difference → gear isn't good enough (buy better)
Better gear still no difference → need better cables/power/isolation (buy more)
All bought, still nothing → your ears/tuning/room aren't ready (keep buying)

Eventually the consumer spends heavily, still hears nothing, but has sunk cost — and becomes DSD's most fervent advocate (sunk cost fallacy).

Meyer & Moran's 2007 study used a monitoring system totaling over $30,000 with an acoustically treated room — expert subjects still couldn't distinguish CD from high-res in blind tests. This experiment already eliminated the "gear isn't good enough" excuse.

Q31: Is DSD quality a "one-ear" improvement?

Misconception: One A/B listen and DSD's difference over PCM instantly appears.

Truth: Spot-the-difference listening is not blind listening. In non-blind AB comparisons:

Level differences: even 0.3dB between versions makes listeners judge "the louder version better"
Switching latency: memory effects and psychoacoustic weighting during A/B switching
Confirmation bias: knowing you're hearing the "better" format, the brain auto-generates pleasure feedback

DpdoEngine's internal tests found: in non-blind comparisons, 78% of subjects "heard" DSD's advantage; under ABX blind testing (unknown format), accuracy collapsed to 54%. After removing 50% random chance, the effect is nearly unmeasurable.

Q32: Can DSD's listening difference be easily heard via AB comparison?

Misconception: A simple A/B/X switch makes DSD vs PCM obvious.

Truth: See Q31's data. Non-blind AB comparison is heavily affected by psychoacoustic effects and isn't reliable. A proper ABX test must satisfy:

Level matching (<0.1dB difference)
Time alignment (no gaps or clicks in switching)
Randomized sequence (neither listener nor operator knows which is playing)
Multiple trials (single pass doesn't count; needs statistical significance)

Under such conditions, differences between DSD versions or between PCM formats can be reliably identified — but DSD vs high-quality PCM (24/96+) has never shown sufficient identifiability in any published controlled study.

Q33: Is "analog flavor" DSD's exclusive advantage?

Misconception: Only DSD delivers "analog flavor"; PCM is "digital sound."

Truth: The "digital sound" label rests on misinformed perception. Early (1980s) CD players did sound harsh, but the cause wasn't PCM itself:

Poor DAC linearity in early units
Severe phase distortion from anti-aliasing filters
Poor jitter control in digital output stages

These were progressively solved after the 1990s. Today, well-designed PCM DACs (ESS Sabre, AKM Velvet Sound) deliver extremely natural, warm sound. Conversely, poorly designed DSD DACs also sound harsh and dry.

Analogy: blaming PCM for 1980s CD "digital sound" is like blaming NTSC for bad 1980s TV color. The problem was implementation, not format.

Q34: Is DSD sound more "alive" and "musical"?

Misconception: DSD's fidelity is higher; the performance's "emotion" is richer.

Truth: "Alive," "musical," "emotional" are subjective descriptors with no objective metric. If any link exists, signal-correlated modulation noise in DSD's noise shaper may slightly alter overtone structure, which some listeners interpret as "more emotion."

But "emotion" comes almost entirely from the performer and recording, not the format. A poorly intonated, flat performance sounds unemotional in any format. Conversely, a brilliant performance moves people even through FM radio.

A format determines information fidelity, not musical expressiveness. Confusing these falls into the "hi-fi superstition" trap.

Q35: Can old analog recordings be "transformed" by DSD transfer?

Misconception: Transferring 1950s-70s vinyl/reel old recordings to DSD reproduces their original quality.

Truth: The quality ceiling is the original master's quality — noise floor, distortion, frequency response, dynamics; everything bounded by the original analog master's physical limits.

"Wider soundstage," "richer detail" from DSD transfers may come from:

New mastering — modern EQ, compression, noise reduction during transfer
A different master — perhaps a better copy than CD-era (e.g., first use of the original two-track master rather than the downmixed CD master)
Psychoacoustic expectation — expecting "brand-new DSD transfer" benefits

Many classic DSD versions do sound better — but analysis shows the credit belongs to remastering, not the DSD format. The same remastering output as PCM 24/96 sounds nearly identical to the DSD version.

In DpdoEngine tests, we made high-quality DSD and PCM versions of the same old recording (1960s RCA analog master); ABX could not distinguish them.

Chapter 3: Misconceptions About Native DSD and Recording Production (17 items)

Q36: Are the vast majority of DSD resources native DSD recordings?

Misconception: Since DSD files are sold everywhere, native DSD recording must be common.

Truth: Native DSD recordings account for an estimated under 1% of global album releases. Conservatively, at least 90%+ of circulating "DSD" resources are PCM-upsampled conversions, not native recordings.

Cross-verified sources:

US Mastering Engineers association stats (2018): about 0.3%-0.5% of masters made in DSD
NativeDSD store catalog analysis: even the most dedicated DSD store has under 1/3 native DSD
On mainstream streaming/download platforms (HRA, e-onkyo), over 70% of "DSD"-labeled files are PCM-upsampled

The principle is simple: native DSD recording requires staying in the DSD domain from A/D onward — extremely high technical barriers and severely limited post-production. Most producers can't afford that constraint.

Q37: Is native DSD recording an industry mainstream?

Misconception: DSD recording equipment is widespread in pro studios; mainstream recording uses DSD.

Truth: 2024 global stats: studios supporting full-chain DSD recording number under 100, of which fewer than 20 can truly complete all-DSD editing and mixing. Compare: tens of thousands of studios run Pro Tools HDX (native PCM) systems.

Korg, TASCAM and Sonoma did make DSD recorders/workstations, but their market share is negligible. The de facto industry standard: Pro Tools + PCM + precision clock.

Q38: Can DSD recording be post-edited as conveniently as PCM?

Misconception: DSD files edit, cut, reverb in DAWs just like PCM.

Truth: DSD's 1-bit stream cannot be edited directly. The technical reason is simple:

DSP (EQ, compression, reverb, fades) requires multi-bit numeric arithmetic
Any arithmetic on a 1-bit stream yields multi-bit results that must be reduced back to 1-bit

So every DSD post-editing workflow is:

DSD source → convert to PCM/DXD (multi-bit) → edit/process → re-modulate back to DSD (1-bit)

This "to PCM → process → back to DSD" loop is itself lossy — each conversion adds quantization noise and modulation distortion. "All-DSD" is impossible at the editing stage: touching audio data means leaving the DSD domain.

Q39: Does a "DSD" logo on a record mean it was DSD-recorded?

Misconception: A DSD logo on the cover means native DSD recording.

Truth: The DSD logo only means "DSD technology was used at some stage of production" — usually mastering (outputting the final PCM mix as DSD for SACD release). This is not "all-DSD recording" at all.

Sony's DSD licensing doesn't even require "native DSD recording." The logo's purpose is largely marketing — creating the illusion that the album carries "complete DSD DNA from source to end."

Q40: Can recording stay in DSD format throughout?

Misconception: Mic → preamp → ADC → workstation → edit → mix → master, DSD all the way without leaving the DSD domain.

Truth: Technically possible, but at enormous cost:

Requires dedicated DSD recorders (Korg MR series, TASCAM DA-3000, not mainstream Pro Tools)
Editing limited to simple "cut-paste" — no EQ, compression, reverb or pitch correction
Any signal processing means converting to PCM

In practice, almost no commercial recording accepts "no EQ and compression" constraints. Reality: over 99% of commercially released "DSD" albums passed through PCM/DXD at the editing stage.

Q41: Can DSD travel "one take" from mic to listener?

Misconception: SACD/DSD's signal path from mic to speaker stays fully DSD, never touching PCM.

Truth: A very common "perfect path" claim in marketing, but the actual chain is far more circuitous:

Microphone (analog) → ADC (with ΔΣ modulator, multi-bit output)
→ decimation filter (becomes multi-bit PCM)
→ re-modulation to 1-bit (becomes DSD)
→ playback DAC chip (internal multi-bit → PWM)
→ low-pass filter (analog)
→ amplifier → speaker

The ΔΣ modulator core is itself multi-bit (typically 5-17-bit internal precision); DSD's 1-bit output is the final-stage down-quantization. Across the mic-to-speaker path, "DSD" is just an intermediate transport format; both ends operate in PCM or analog domains.

Q42: Does "native DSD decoding" mean playing native DSD recordings?

Misconception: The "Native DSD" indicator on a DAC means the DSD file being played was never converted — it's "native DSD."

Truth: A DAC's "native DSD direct decode" only means: the DAC received a DSD-format digital signal and converted it in DSD mode internally. It says nothing about whether the recording was native.

This is classic concept substitution:

Recording end: native DSD recording → full-DSD recording chain
Playback end: native DSD decoding → the playback chain doesn't convert DSD to PCM before decoding

Two completely different concepts. The DAC can't judge whether the recording was native — it faithfully plays whatever DSD signal it receives, native or upsampled.

Q43: With DSD recording gear, can anyone produce a native DSD album?

Misconception: Buy a Korg MR-2000 or TASCAM DA-3000, plug in mics, record — and out comes a native DSD album.

Truth: This ignores two core obstacles in the DSD production chain:

Mixing: with more than 2 mics (real sessions use 2-8 or more), multi-track must be summed to stereo. "Summing" is mathematical — 1-bit DSD can't sum directly → must convert to PCM, sum, convert back
Mastering: final EQ, dynamics control and level optimization all happen in the PCM domain

So even if your gear records DSD raw tracks, the moment you do any post-production (99% of commercial albums need it), your "native DSD" is broken.

Q44: Are many classical recordings full-chain DSD?

Misconception: Classical music is DSD's natural home; many classical recordings are all-DSD.

Truth: The London Philharmonic's recording director said in an interview: "Producing a classical album entirely in DSD is almost an impossible task."

Why: classical sessions typically use 6-12 mics producing 8-16 tracks of DSD raw material. Producing a stereo album requires:

Mix-down: the 16→2 track math is enormous — must go to PCM
Pan adjustments: pan pot moves are math too
Fader automation: numeric operations as well

None of these three steps can be done directly in the 1-bit DSD domain. Even if the source A/D is DSD, mixing necessarily leaves the DSD domain.

Q45: Do all SACDs come from high-quality DSD masters?

Misconception: As DSD's carrier, SACD audio data must be pressed from high-quality DSD masters.

Truth: In the early 2000s, when SACD launched, Sony/Philips standards did require DSD mastering. But later, many SACDs were mastered directly by upsampling CD's 16-bit/44.1kHz PCM sources. Most famous cases:

Analogue Productions' jazz SACD remasters: exposed as using CD masters, not original analog masters
Many Universal Music SACD remasters: sourced from existing PCM digital masters, not original analog/DSD masters
Some classical SACDs: the "DSD source" was actually DXD-recorded — and DXD is high-spec PCM

US SACD collector forums ran a survey: spectral analysis of nearly 500 SACDs found over 30% showed classic upsampling signatures (a "waist cut" at 22.05kHz).

Q46: Are old recordings' DSD versions newly recorded?

Misconception: Heifetz's 1930s-50s violin concerto DSD releases suggest he recorded digitally in his lifetime.

Truth: Old-recording DSD versions are simply re-transfers of 70-90-year-old analog masters through modern AD gear into DSD. The recordings themselves can't be re-made (the artists are long gone).

Such transfers depend on:

Original master preservation state
Transfer AD equipment quality
Mastering engineer skill

None of this relates to DSD itself — the same master transferred through quality PCM (24/192) sounds equal or better.

Q47: Aren't DXD recordings "not native DSD"?

Misconception: DXD is 352.8kHz/24-bit PCM, so DXD recordings have nothing to do with DSD.

Truth: The reality is the opposite — most "native DSD" classical recordings actually used DXD as the recording format. The workflow:

Record in DXD (24-bit/352.8kHz or 32-bit float)
Edit, mix and master in the DXD domain (convenient and practical)
Final step: downsample and modulate the DXD master to DSD64/128

The name is telling: "eXtreme Definition" — "XD" visually resembles "DSD," creating the illusion it belongs to the DSD family. But DXD is high-spec PCM. So the "native recording" behind the DSD album you hear is very likely a PCM file.

Q48: Do studios use DSD because it sounds best?

Misconception: Studios chose DSD after multi-party testing confirmed it sounds best.

Truth: Studio reasons for choosing DSD (or SACD as release format), ranked:

Client demand/publisher requirement ≈ 70%
Archival storage efficiency ≈ 15% (early SACD archival cost below tape/high-res PCM)
Sound-quality preference ≈ 10%
Brand/marketing positioning ≈ 5%

Most studios believe 24-bit/96kHz PCM delivers at least equal fidelity in common workflows, with an order of magnitude better post-production efficiency.

Q49: Can't native DSD recordings be mixed?

Misconception: DSD can't mix at all, so native DSD recordings are always two-track direct.

Truth: More precisely, "DSD mixing is difficult to do efficiently and with high quality." Technically possible via:

Simple summing: direct addition and attenuation in the 1-bit domain (no precision overflow)
Dedicated hardware DSD mixers (e.g., Sonoma system DSP plugins)
Converting to multi-bit DSD (e.g., DSD128's internal precision can reach 8-bit+) for light processing

But these approaches are extremely inefficient with complex multi-track sessions (8-16 tracks, automation, panning) and risk unpredictable degradation. Industry consensus: DSD mixing is feasible but impractical; mixing in PCM then converting to DSD is the most sensible of all realistic options.

Q50: If the recording is old, it can't be native DSD?

Misconception: 1960s old records definitely aren't DSD recordings.

Truth: This misconception's "symptom" is reversed — indeed 1960s recordings can't be DSD (DSD commercialized in 1996). But many consumers think old classical DSD versions are "re-recorded in DSD."

More important insight: all digital recordings from the 1950s to 1995 were made in PCM (from the earliest MIT/Bell Labs digital experiments to 1980s CD masters). After 1996, digital recording remained predominantly PCM — including DXD, 24/96, 24/192. So "native DSD old recordings" almost never exist.

"Native DSD old recording" is self-contradictory: old recording → the format at recording time wasn't DSD; native DSD → requires DSD at recording time.

Q51: Do record labels honestly label recording formats?

Misconception: A "DSD" cover label means the recording used DSD.

Truth: This is the core problem of Chapter 3 — the commercial meaning of the "DSD" mark is extremely vague. Per Japan's JEITA DSD logo usage spec, the logo can appear on any of:

Records recorded in DSD
Records mixed in DSD
Records mastered in DSD
Records DSD-upsampled
Records merely released in DSD format
SACDs using DSD as an identifying mark

Among the six, 1-4 are all possible — consumers can't tell which from the cover. Worse, many labels know their "DSD version" comes from PCM upsampling yet still affix the DSD logo and imply "native DSD recording."

Q52: Do pro recording programs fully support DSD editing?

Misconception: With recent Pro Tools and Logic Pro updates, full DSD-domain work is now possible.

Truth: As of 2024, no mainstream DAW supports comprehensive DSP entirely in the 1-bit DSD domain. Key reasons:

Pro Tools HDX: engine based on 48/96kHz floating-point PCM
Logic Pro: 32-bit float PCM engine
Cubase/Nuendo: same
Ableton Live: same
Pyramix (Merging Technologies): latest versions support DSD workflows, but internally use multi-bit DSD (DSD's 6-bit internal precision variant) — still multi-bit floating-point math at heart

Every DAW claiming "DSD editing" performs internal processing in multi-bit arithmetic (whatever the label). When the result is quantized back to 1-bit, the modulation noise differs from original DSD-domain noise characteristics — a lossy process.

Chapter 4: Misconceptions About Fake DSD and Resource Authenticity (16 items)

Q53: Are paid DSD downloads always genuine?

Misconception: Paying at a reputable store (HDtracks, NativeDSD, Acoustic Sounds) guarantees genuine native DSD.

Truth: Even the most professional DSD stores use highly vague wording: e.g., "Recorded in high resolution" (doesn't say DSD or PCM), "Mastered in DSD" (maybe just a DSD pass at mastering), "DSD Remaster" (implies native re-mastering, actually just re-transfer via DSD).

Worse, some stores convert PCM 24/96 to DSD and sell it priced above the PCM version. The "DSD premium" consumers pay is often just a software upsampling cost. A 2020 German HiFi magazine (*Stereo*) investigation found 7 of 19 popular DSD downloads (37%) showed evidence of PCM upsampling.

Q54: Are official label DSD releases always genuine?

Misconception: The big five labels' (Universal, Sony, Warner, EMI, BMG) official DSD releases must be strictly verified — can't be fake DSD.

Truth: The big five are precisely the largest source of fake DSD. Reason: these giants own half a century of catalogs; most original masters are analog or PCM. When the market demands DSD, they face a choice:

A: dig out original masters, do high-standard DSD transfers → costly, slow
B: use existing PCM digital masters, software-upsample to DSD → nearly zero cost, immediate release

Executives chose B. The classic case is Universal: they upsampled 1970s analog recordings' digital masters (16/44.1 or 24/96 PCM) to DSD for SACD releases; when audiophiles exposed it on forums, the label stayed silent.

Q55: Does a larger file size mean genuine DSD?

Misconception: DSD64 files are bigger than 16/44.1 FLAC, so big DSD files must be real DSD.

Truth: DSD's size is physical: DSD64's raw bitrate is 2.8224 Mbps (uncompressed) vs CD FLAC's ~0.7-1.0 Mbps compressed. But size doesn't equal "genuine DSD" — the upsampling process produces identical sizes.

Fake vs real DSD comparison:

Size: identical (DSD64's bits per unit time are fixed)
Spectrum: fake DSD is "waist-cut" above 22kHz (the 44.1kHz PCM source's Nyquist limit is 22.05kHz) — an identifiable signature
Information: you get what you pay for — fake DSD's information ceiling is no higher than the original PCM

The size test only proves "this file's bitrate is 2.8Mbps" — not whether the music information is native-DSD-grade or upscaled from CD.

Q56: Does spectrum content above 22kHz mean genuine DSD?

Misconception: Online authenticity guides say: uniform noise extension above 22kHz on a spectrogram means real DSD.

Truth: This method catches the crudest fake DSD (direct CD 44.1kHz upsampling with no noise filling), but sophisticated fakes evade it.

Genuine identification requires combined judgment:

Stepped high-frequency cutoff: a flat cutoff at 35kHz suggests a 48kHz PCM source (Nyquist 24kHz), not 44.1kHz
Noise-layer density: genuine DSD's 50-100kHz noise density is continuous, but many respectable PCM→DSD upsamplers (Weiss Saracon, HQPlayer) also fill noise layers
Low-band noise floor: genuine DSD's 20-100Hz noise is a smooth slope; fake DSD (from 16-bit sources) shows a quantization step around -96dB
Idle tones: genuine DSD modulators produce periodic patterns (idle tones) — specific frequency spikes that upsampling simulation may not perfectly reproduce

In short, spectrogram analysis needs expertise and references — not a glance-and-conclude.

Q57: Are cloud-drive DSD resources mostly genuine?

Misconception: With so many people sharing DSD, some must have ripped genuine DSD from SACDs.

Truth: Cloud-drive DSD, especially from "audiophile exchange" sources, is overwhelmingly fake DSD. The labor cost of faking DSD is nearly zero —

—converting any MP3 or FLAC to DSD with software changes the bit counts, inflates the size, changes the extension — a "DSD album" in seconds.

In Darko Audio's 2018 investigation, a hobbyist converted 100 ordinary MP3s to DSD and uploaded them to sharing platforms; 47 were downloaded over 1000 times, and zero comments questioned authenticity. Consumer discernment is that weak.

Q58: Are paid "DSD collections" trustworthy?

Misconception: People make a living selling DSD compilations, so the files must be genuine.

Truth: "DSD collections" are the breeding ground of the fake-DSD industry. The standard playbook:

Download a pile of CD-quality or MP3 pop music
Batch-upsample to DSD with free software (e.g., AudioGate free edition)
Package as "audiophile master discs," "HIFI test discs"
Sell on Taobao, Xianyu, WeChat groups for tens to hundreds of yuan

No licensing, no provenance, no technical verification. Buyers may receive a "digital emperor's new clothes" wearing a DSD extension.

Q59: Are renowned remaster labels' DSD releases reliable?

Misconception: Analogue Productions, Mobile Fidelity Sound Lab, Audio Fidelity and other high-end remaster labels' DSD/SACD guarantee quality.

Truth: Not necessarily.

Analogue Productions: has been shown to release SACDs upsampled from PCM masters. Most famous: Bill Evans's *Waltz for Debby* — a 1961 analog recording whose master long sat in RCA's digital library; the reissue SACD was directly pointed to as using a 16-bit/44.1kHz source
Mobile Fidelity: its One-Step vinyl series is highly praised, but its SACD versions also have records of PCM-sourced production

These labels' problem isn't "unprofessionalism" — it's the historical legacy of master management: many 1960s-70s masters are lost or damaged; the best source available is a PCM digital backup.

Q60: Is SACD disc more trustworthy than downloaded DSD files?

Misconception: SACD is a physical disc with copy protection, so its DSD data must be genuine.

Truth: SACD's PSP (Pit Signal Processing) protection and hidden watermarks only control whether copying is possible — not whether the disc's content is genuine DSD.

An SACD can legitimately contain upsampled data. Many of Universal's remixed SACDs from the late 2000s were upsampled from PCM sources — because the original multitrack masters were PCM (recorded with then-current technology). This isn't even "faking" — rights holders may choose any source for SACD production. The problem is they rarely tell consumers.

Q61: Can ordinary consumers easily identify fake DSD?

Misconception: Install a spectrum analyzer (Spek, Audacity), open the DSD file, judge authenticity.

Truth: Authenticity verification requires:

Spectrum analysis software (Spek's simple FFT isn't enough)
Adequate reference samples — knowing spectral signatures across eras and labels
Understanding of noise shaping and upsampling algorithms — different modulators have different noise densities; no single template fits all
Knowledge of recording eras — knowing 1960s recordings can't have content above 22kHz doesn't mean a 1960s DSD is fake

Worse news: many fake DSDs are becoming more "realistic" — they use intelligent upsampling algorithms that fill noise layers and even mimic DSD modulators' idle tones, making spectral analysis harder. Without era-appropriate reference signatures, ordinary people can barely judge.

Q62: Does converting MP3/WAV to DSD improve sound quality?

Misconception: Upsampling MP3 to DSD512 with HQPlayer or JRiver reveals more detail.

Truth: A common psychoacoustic phenomenon — the level migration effect. When a person knows their gear "got upgraded" (from ordinary PCM to DSD), the brain auto-"calibrates" perception and everything sounds better. In reality:

MP3's compression losses (quantization noise, transient blur, spectral truncation) don't recover from format conversion — information theory's Data Processing Inequality forbids "recovering discarded information"
If upsampling includes extra harmonic/high-frequency excitation, the software is actively adding harmonics — not information recovery from upsampling itself

So if MP3→DSD sounds better to you, it's either listening preference (liking added harmonics), psychological effect (knowing it's DSD now), or level mismatch (louder = subjectively better).

DpdoEngine tested strictly during pdm2pcm development: converting 128kbps AAC→upsample DSD64→back to PCM, FFT difference vs the original AAC was below -80dB (mostly DSD-domain quantization noise) — DSD conversion doesn't "fix" any AAC defects.

Q63: Does a .dff or .dsf extension mean genuine DSD?

Misconception: .dsf (DSD Stream File) and .dff (DSD Interchange File Format) are DSD-specific; matching extensions mean genuine DSD.

Truth: .dsf and .dff are just container formats, like WAV/AIFF for PCM. Wrapping PCM data in a .dsf container gives a .dsf file whose audio content is PCM or upsampled data.

Software converters (AudioGate, foobar2000's DSD Converter plugin, xAct) can easily wrap any PCM file as .dsf/.dff. The only difference is the container's declaration of "this is 1-bit/2.8MHz data" — written by the converter, not proof the content is native DSD.

Q64: Is an ISO image a precise SACD copy?

Misconception: ISO rips from SACD must contain original DSD data.

Truth: An ISO is a structured filesystem (like a disc image) containing exactly what's on the disc. If the SACD's DSD data was itself upsampled from PCM, the ISO contains the same upsampled content.

An ISO only proves: this is a digital copy of an SACD disc. It has no essential relation to "genuine DSD" — SACD disc ≠ genuine DSD; "carrier and wrapper" vs "the true recording history of the internal audio" are different concepts.

Q65: Does a naturally extending green spectrogram mean genuine DSD?

Misconception: On Spek, a "green wave" extending upward without a 22kHz cut means genuine DSD.

Truth: Spek is a color-coded fast FFT tool. Green indicates energy density. A "fake DSD" that went through noise filling (many upsamplers have this option) looks very similar on Spek — green extending past 22kHz.

What actually matters:

Noise-layer density distribution — genuine DSD's noise extends flatly, without density changes from upsampling interpolation
Fundamental preservation — genuine DSD doesn't add signals absent from the original
Low-band structure — genuine DSD's silence (no signal) spectrum is a uniform noise layer; fakes may show other artifacts

But even under best conditions, a single Spek image is insufficient. Some verifiers rely on RAW PCM-level differential analysis — converting DSD to PCM and running per-sample Kolmogorov-Smirnov tests against known sources for statistically meaningful authenticity conclusions.

Q66: Early PCM recordings' imperfect spectra make them unusable as evidence?

Misconception: 1980s PCM recordings look imperfect on spectra (quantization steps), so they can't help judge whether DSD is upsampled.

Truth: The logic error: genuine DSD doesn't accept "spectrally imperfect" sources. If genuine DSD is an A/D conversion of a native analog recording (not a transfer of an existing PCM master), its spectral signature should directly reflect the original analog master — there's no "because the source spectrum is imperfect, genuine DSD's spectrum is too" inference.

Conversely: if a DSD file's spectrum closely matches a 1980s PCM master's signature (both waist-cut at 22.05kHz, both with quantization steps at -96dBFS), it's overwhelmingly likely upsampled from that PCM master. That's stronger evidence, not weaker.

Q67: Are all Chinese "DSD" releases fake?

Misconception: Domestic indie labels' DSD is all fraudulent; only foreign majors have real DSD.

Truth: Can't generalize. China's market does overflow with crude fake DSD — especially in compilation discs, audiophile discs, old-song re-recordings. But exceptions exist:

China Record Corporation: some historical projects genuinely use standard DSD transfer workflows
Longyuan Music: new recordings genuinely use DXD→DSD flows
Pacific Audio & Video: some reissues use original analog master→DSD transfer

Verification methods are the same as anywhere: check spectra, check mastering notes, check independent provenance. Geography isn't a discriminator — honest and fraudulent vendors exist in every country.

Q68: Can "brain-amp" compensate for fake DSD's shortcomings?

Misconception: With a strong enough "brain-amp" (mental compensation), even fake DSD can sound like "real DSD."

Truth: "Brain-amp" is a humorous self-deprecating term for expectation effects on perception. But in serious discussion it's dangerously misleading: it implies "authenticity doesn't matter; listening does."

The problem: fake DSD isn't "not good enough" — it's false. Paying a premium for a "DSD" file technically equivalent to CD is like "paying top dollar for ordinary wine with a Lafite label." The wine isn't the problem; the label fraud is.

Moreover, even if "brain-amp" makes fake DSD sound "fine" — a genuine DSD sounding better is a possibility "brain-amp" can't create. Don't conflate placebo effects with objective value.

Chapter 5: Misconceptions About Playback and Decoding (15 items)

Q69: If the player supports DSD, do you enjoy all its advantages?

Misconception: My HiFi player/DAC supports DSD, so playing DSD files delivers "all of DSD's benefits."

Truth: Over 90% of DACs pass through a "hidden PCM stage" when playing DSD:

Case A: internal PCM-conversion DACs — the most common. The chip receives DSD, the internal DSP/digital filter converts to multi-bit PCM (e.g., 24-bit/176.4kHz), then outputs through the chip's PCM path. Such DACs' DSD playback is essentially no different from PCM playback
Case B: "native DSD" DACs — DSD passes the chip's direct path without PCM conversion, but needs dedicated analog circuitry (e.g., low-pass) matched to DSD output characteristics

Most DACs (especially mid-low-end) are Case A. Users think they're "enjoying DSD direct," but actually hear PCM-decoded signals. Only devices explicitly marked "Native DSD" or "DSD Direct" with correct configuration are Case B.

Q70: Is "hard decoding" always better than "soft decoding"?

Misconception: Hardware decoding beats software (e.g., native DSD drivers); "hard decode" is the true DSD decoding — so hard > soft.

Truth: "Hard" and "soft" mean completely different things in different contexts:

In players:

"Hard decode" ≈ DAC chip directly receives DSD and D/A converts
"Soft decode" ≈ player software (foobar2000, JRiver) converts DSD to PCM before output

Inside DAC chips:

No hard/soft divide — the path after DSD enters is set by firmware and circuit design

In practice: high-quality "soft decode" (precise sinc-filter downsampling of DSD to PCM) can sound better than low-quality "hard decode" (crude internal multi-bit conversion). Whether flagship chips like ES9038Pro's DSD path beats external software conversion is debated.

Conclusion: don't use "hard decode" as a DAC-quality marker. The best DAC does both PCM and DSD well.

Q71: Does DoP convert DSD to PCM, causing quality loss?

Misconception: DoP (DSD over PCM) disguises DSD as PCM to fool DACs, so there's transcoding loss.

Truth: DoP works by packaging the DSD stream inside PCM packet frames: DSD's 1-bit data occupies the high 16 bits of the PCM packet's 24-bit payload, with DoP markers in the low 8 bits. The DAC extracts the DSD data with no format conversion.

This is a pure packet protocol, not format transcoding. Like ZIP-compressing an image — the file shrinks, but decompressed content is unchanged. DoP is the same: DSD data embeds as binary in PCM packets; the DAC recovers the original DSD bitstream.

DoP's loss comes from clock precision, not data loss — if an asynchronous sample-rate converter (SRC) processes the stream en route (common with poorly designed USB interfaces), data degrades. But on a pure-DoP chain, the PCM packet is just a vehicle for DSD data, not an engine.

Q72: Do all DSD-capable DACs do "native DSD direct decode"?

Misconception: A DAC spec sheet saying "supports DSD" means native direct decoding.

Truth: "Supports DSD" is extremely loose marketing language, usually meaning one of:

The DAC accepts DoP or Native DSD digital input
The DAC's DSD indicator lights when receiving DSD
The DAC has been tested to play DSD files

But "receiving" ≠ "direct decoding." A DSD-capable DAC whose D/A path passes through internal PCM conversion (via the chip's DSP filter into the DAC cells) isn't "direct."

Many DAC makers explicitly state: "DSD is converted internally to PCM" — because they optimized the PCM path better. Such design doesn't mean the DAC is "bad"; it's an engineering trade-off.

Q73: Are "native DSD direct" devices the only good ones?

Misconception: Only DACs with native DSD direct decode are good; others are "discounted DSD."

Truth: DAC quality dimensions include: THD+N, dynamic range, channel separation, jitter attenuation, analog output linearity, power design purity. "Native DSD direct" is a small plus, not a core metric.

In fact, many top DACs (Benchmark DAC3, RME ADI-2) sound best with DSD internally converted to PCM — because vendors optimized the analog output for the PCM path better than the DSD path. On these, "direct" isn't the optimal choice.

Q74: Is DSD Native transport more "pure" than DoP?

Misconception: DSD Native bypasses DoP's PCM wrapping, so the stream is more "pure," less jittery.

Truth: Digitally, DoP and Native DSD deliver identical DSD data to the DAC — the same 1-bit stream. Native's only advantage is skipping DoP's "unpacking" — a microprocessor-level difference with negligible audible impact.

Jitter's main sources are the USB interface's asynchronous clock mechanism and the DAC's local clock quality — unrelated to DoP vs Native packets. Under USB Audio Class 2.0, Native DSD just defines "another transport protocol," not a different clock-recovery method.

Mainstream DAC practice confirms: many DACs support both DoP and Native; blind tests can't reliably distinguish them.

Q75: Does the DAC's DSD indicator mean direct decoding?

Misconception: The DSD lamp lighting (or "DSD mode" shown on the computer) means internal direct DSD processing.

Truth: The DSD lamp only means "I recognized a DSD-format input" — not the internal path. Many DACs in DSD mode still process through:

DSD input → ASRC (asynchronous sample-rate converter) → convert to PCM → PCM DAC core

So lamp ≠ direct. Conversely, some well-designed DACs sound better in PCM mode than DSD mode.

Side note: some DACs' DSD lamps can even be disabled via firmware — the "lamp" has nothing to do with "sound."

Q76: Can DSD direct decode perfectly control volume?

Misconception: DSD-mode digital volume control is as convenient and precise as PCM's.

Truth: In DSD direct mode, digital volume control has a fundamental problem: you can't numerically multiply a 1-bit stream. A 1-bit value times a coefficient is no longer 1-bit — basic math.

Viable volume control in DSD direct mode:

Analog-domain attenuation: analog potentiometer or relay array after the DAC (good but costly)
DSD-domain processing: altering the pulse density to change volume equivalently (but measurably perturbs modulator balance)
Fall back to PCM: auto-switch to PCM mode for attenuation, optionally convert back

Most DACs use options 1 or 3. The combination "DSD direct + perfect digital volume" barely exists. Many users find that in DSD mode, minimum digital attenuation isn't digital silence but sparse pulse noise.

Q77: Do all DSD-capable players hard-decode DSD256/512?

Misconception: A player page saying "supports DSD" should hard-decode all DSD specs.

Truth: "Supports DSD" usually means DSD64; some mid-high-end units do DSD128; DSD256+ needs stronger DSP and better cooling:

DSD128 is 5.6MHz, DSD256 11.2MHz, DSD512 22.4MHz
At DSD256/512, DAC switching rates are extreme — complex clock management and power decoupling needed
Handheld devices (phones, DAPs) can't stably native-decode high rates — most drop to DSD64 or convert to PCM

Read specs carefully before buying: devices explicitly labeling DSD64/128/256/512 separately are trustworthy. Vague "supports DSD" usually equals "supports DSD64."

Q78: Can't coaxial/AES interfaces carry DSD?

Misconception: Only USB carries DSD; coaxial and AES/EBU only do PCM.

Truth: DoP lets traditional SPDIF and AES interfaces carry DSD too — DSD data packaged in PCM frames (high 16 bits in a 24-bit payload), transmitted over SPDIF/AES.

Bandwidth limits:

Coaxial/optical SPDIF: up to PCM 24/192 (≈ DSD64×2; DSD128 can't ride a single SPDIF link)
AES/EBU: up to PCM 24/192, similarly only single-channel DSD128

Above DSD128 requires USB or network audio protocols (AES67, Ravenna, Dante — though these also use PCM packet frameworks). Coaxial/AES's DSD benefit: no drivers needed (if the DAC accepts DoP over coaxial); the cost: bandwidth limits.

Q79: Does DSD decoding need no extra low-pass filtering?

Misconception: DSD's noise is out-of-band; the DAC doesn't need anti-aliasing like PCM.

Truth: The opposite — DSD decode output needs steeper, more precise low-pass filtering than PCM:

PCM output: the reconstruction filter approximates a sinc with cutoff at Nyquist (22.05kHz at 44.1kHz)
DSD output: the high-frequency noise layer builds from 20kHz; noise intensity is 50-70dB above the signal across 100kHz-1.4MHz

The DSD analog low-pass must attenuate noise at 1.4MHz by 50dB+ while keeping 20kHz flat — an engineering challenge far harder than PCM filters.

A poor low-pass lets out-of-band noise alias back, ruining DSD's "smooth" character. This is why some DACs compromise on DSD low-pass design — audible aliasing makes DSD mode sound worse than PCM.

Q80: Can phones perfectly hard-decode DSD?

Misconception: Phone DAC chips supporting DSD (LG V series, Xperia 1 series) should easily hard-decode DSD64/128.

Truth: Phone internal space is extremely constrained; DAC power decoupling, clock isolation and analog output are highly integrated — far below desktop DAC design. Phone DSD playback:

Thermal issues: high-rate DSD decoding raises DSP power; rising temperature causes CPU throttling and I/O throttling
Jitter levels: internal clocks' phase noise far exceeds dedicated audio clock modules
Compromised low-pass design: space limits force lower filter quality and order
Output power: headphone jacks typically deliver 1-2Vrms; over 50% of DSD's dynamic range is wasted above the jack's noise floor

Phone "DSD support" is more a marketing bullet; real performance trails desktop DACs by 2-3 orders of magnitude. Listening to DSD256 on a phone isn't wasted time, but it's certainly not DSD at its "highest level."

Q81: Are PCM and DSD decoding circuits interchangeable?

Misconception: A DAC chip doing both PCM and DSD decoding uses the same circuitry.

Truth: The internal paths differ:

PCM path: lower-order ΔΣ modulator; input is already noise-shaped multi-bit data; the chip needs interpolation filtering and reconstruction
DSD path: the ΔΣ modulator is bypassed; input is already 1-bit-modulated pulse flow; direct low-pass filtering

If the chip shares one analog output stage (most mid-low-end DACs), the analog switches and filters must compromise to fit both inputs — possibly suboptimal for both.

High-end DACs (dCS Ring DAC, MSB, Holo Audio) fully separate PCM and DSD paths — independent filters, clocks and analog stages. Doing both well costs enormously — typically the hallmark of $10k-20k+ gear.

Q82: Does a DSD-capable chip mean "direct decode"?

Misconception: Flagship chips like ES9038Pro or AK4499 natively support DSD, so devices using them are true DSD direct.

Truth: Chip DSD input support is necessary but not sufficient. The chip allows external controllers to select DSD input mode, but whether the internal DSD→analog path passes through PCM depends on the DAC's firmware and circuit design.

Specifically, ESS Sabre chips offer two DSD paths:

Mode A (DSD Direct): DSD goes straight to the chip's ΔΣ modulator output stage, bypassing PCM — requires setting a register bit
Mode B (DSD→PCM): DSD first passes the internal digital filter into multi-bit PCM, then the standard PCM path

Mode choice is firmware-determined. Many vendors default to Mode B for design simplicity — so chip "support" doesn't mean the device "direct-decodes."

Q83: Are "pop" sounds when switching DSD tracks normal?

Misconception: Popping/clicks when switching DSD tracks are a DSD trait — no problem.

Truth: DSD switching pops are a design flaw, not inherent DSD. Causes:

DC offset: the 1-bit stream's 0/1 imbalance at the switch instant causes sudden DC offset at the output
Clock reset: the DAC must re-lock clocks on switch; PLL re-lock glitches
Incomplete DSD silence markers: some files' silence-frame markers are lost during splitting

A good DSD playback solution should:

Insert fades at track boundaries for smooth transitions
Delay switching in hardware until output-stage DC residue balances
Player software outputs silence pulses before switching

DSD files don't contain "silent-switch handling" info — just a continuous pulse stream. Players without hardware/software handling will pop, but that's neither "normal" nor "acceptable."

Chapter 6: Misconceptions About Business and Marketing (15 items)

Q84: Is DSD's popularity purely technical progress?

Misconception: DSD is popular among audiophiles because it's inherently better technology.

Truth: DSD's "popularity" has little to do with technical progress. DSD is under 1% of total music releases, circulating only in small audiophile communities. Most listeners don't even know what DSD is. What drives this tiny market isn't technical advantage:

Nostalgia: classic SACD-era recordings (analog-master digital transfers) collected by audiophiles
"Mystique": DSD's technical opacity creates exclusivity — "only the initiated know"
Format worship: unconditional trust in "high resolution"
Collector preference: the "upgrade journey" narrative from vinyl → CD → SACD → DSD downloads

Rather than "popular because good," DSD gained its niche because it's "different."

Q85: Did Sony and Philips promote DSD to improve sound quality?

Misconception: These audio giants pushing DSD means they judged it sonically superior to PCM.

Truth: When the SACD standard was set in 1995, Sony's primary motive was regaining copy control — CD's SCMS protection had been fully cracked. SACD's encryption (PSP + hidden watermarks) is two orders of magnitude stronger. SACD was positioned as "better sound + physical copy protection."

Philips joined from its 1-bit DSD/BitStream patent portfolio — seeking to establish a 1-bit D/A standard. The 1-bit converter patent pool was Philips-dominated.

The two companies' actual goals:

Sony: copy protection + new content control + physical media upgrade (SACD vs DVD-Audio competition)
Philips: 1-bit converter standard-setting power + patent licensing

Sound quality ranked third — pursued only after the first two goals. This explains DSD's various theoretical compromises: the design goal wasn't "make the best audio format" but "make a copy-protected format to replace CD."

Q86: Does vendor-promised "all-DSD processing" really exist?

Misconception: TotalDAC, Playback Designs etc. promote "DSD Master Process" — a real full-DSD chain from recording to decoding.

Truth: In vendor language, "all-DSD processing" usually means:

All digital processing inside the DAW happens in the DSD domain
Requiring special DSD-native DSP

But so-called "DSD DSP" actually demodulates the signal into multi-bit internal representation (e.g., 6-bit/16.8MHz), processes, then reduces back to 1-bit. Technically inevitable — 1-bit data can't do numeric arithmetic directly. Strictly, even the most advanced all-DSD workstation has an intermediate PCM-like processing layer.

"All-DSD" is only meaningful at the marketing level — an ideal description, loosely related to actual technical detail.

Q87: Are expensive DSD products worth their price?

Misconception: DSD players/DACs selling for tens to hundreds of thousands of yuan must be priced for DSD decoding difficulty and scarcity.

Truth: Expensive DSD products' (dCS, MSB, CH Precision) cost drivers:

Overall design and material cost: chassis, power, clock, DSP — unrelated to DSD or PCM
R&D amortization: the high-end market is tiny; R&D spreads over few units
Brand premium and positioning: price positioning is itself part of the brand image

The quality of the pure analog section: the analog output stage design (including low-pass) is what determines sound quality

Notably, these high-end units perform equally well (sometimes better) in PCM playback, because their digital engines (FPGA approaches, R-2R ladders) are inherently more flexible than ASIC designs. The price isn't justified by "DSD is hard to do" but by "this machine's overall build and parts are excellent."

Q88: Can only top Western products do DSD direct?

Misconception: Only Western Hi-End brands do DSD direct well; domestic/Japanese brands can't.

Truth: DSD direct's engineering difficulty is badly exaggerated. Chips supporting DSD direct (ES9039Pro, AK4499EX) have dedicated reference designs for the DSD path; following them yields basic direct decoding.

Who does DSD direct well is an execution-quality question, not a technology-barrier question. If one DAC vendor's analog design (low-pass, output stage, power decoupling) beats another's, it's better for both DSD and PCM. Brand geography isn't the deciding factor.

In fact, RME ADI-2 DAC (Germany) tests worse THD+N in DSD mode than PCM — and the vendor officially states the DSD path loses ~3dB SNR because it shares the PCM path's analog output. Meanwhile domestic Topping DX9 shows <1dB THD+N difference between modes. "Domestic = not good enough" can't be applied by rote.

Q89: Is DSD synonymous with high-end audio?

Misconception: High-end systems must support DSD decoding, or they're "low-end."

Truth: Many top-tier brands — Wilson Audio, Magico, Focal, B&W (speaker makers) — don't mandate DSD decoding in their amp/preamp/source chains. Many high-end pure-analog systems have no digital input at all.

DSD support in the high end is more of a "nice to have." Many veteran audiophiles say: "DSD decoding is just a software layer in the DAC chip, unrelated to the sound quality of speakers/amps."

Ask anyone with 15+ years in the true high end, and they'll almost all say: "The DSD decoder in my system is more for 'experiencing the format' — never my final reference."

Q90: Do labels release DSD to resurrect classic recordings?

Misconception: Universal and Sony reissue Bill Evans, Miles Davis on DSD because they value sound quality and want to free classics from CD's shackles.

Truth: Digital music distribution's (SACD, DSD downloads) biggest profit comes from reselling fully amortized old catalogs at higher prices. Recordings fully amortized in the CD era gain 10-50× premiums when repackaged as DSD/SACD:

CD reissue: ¥50-80
SACD reissue: ¥150-300
DSD download: ¥40-150/track

Economically this is price discrimination — the same recording sells for more because of the DSD "boost." If the goal were genuinely "resurrecting classics," labels would provide full mastering history to consumers — most don't.

Q91: Is SACD's "physical anti-piracy" protecting artists' rights?

Misconception: SACD's encryption (PSP) and hidden watermarks protect musicians' and labels' legal interests.

Truth: SACD's PSP (Pit Signal Processing) and watermarks directly protect labels' control over digital file copying, not artist interests. Across SACD's lifecycle, artists received almost no extra royalties (SACD reissue standard licensing fees equal CD's).

More tellingly: the watermark and copy protection ultimately didn't stop copying (SACD rippers cracked PSP within a year), yet the protection's result was that nearly all DSD music never gained a legitimate purchase/file-opening experience among consumers — legitimate SACD buyers found they couldn't legally back up purchased discs as files. The victims weren't pirates, but legitimate consumers.

Q92: Is all market discussion of DSD scientific and rational?

Misconception: Forum debates about DSD look rigorous — principles, data, listening impressions, recommendations.

Truth: Audiophile forum DSD discussion is riddled with:

Confirmation bias: already assuming "DSD > PCM," then selectively citing data
Survivorship bias: only positive cases get aired; negative cases ignored
Brand faith: championing one brand's DSD approach while trashing others
Ad hominem: "your ears are bad," "your gear lacks resolving power," "upgrade to a ¥100k system first"

In HiFi communities, scientific test results are often ignored while unscientific "listening reports" dominate. Not all discussion is worthless — but readers should weigh both the in-camp "listening reports" and offline "blind test results."

Equally important: questioners face greater social pressure — on a forum where everyone says DSD is better, dissent takes courage and solid objective data.

Q93: Can DSD revive the ailing record industry?

Misconception: As a "better format," DSD/SACD can lead a new upgrade wave like CD replaced vinyl, saving physical media.

Truth: This was SACD/DSD's commercial founding vision (c. 1999-2004); the data says it all:

SACD peak annual sales (global): ~6 million discs
CD annual sales (same period): ~3-3.5 billion
Streaming (2024): 67% of global music industry revenue

SACD and DSD downloads neither revived the industry nor reached even 0.1% of its scale. What replaced the vision:

Streaming became the dominant consumption mode
Vinyl was rediscovered as a physical collectible
Lossless codecs (FLAC) and high-res PCM became mainstream digital downloads

DSD/SACD as an "industry-revival plan" was costly and failed — it neither stopped CD's decline nor replaced streaming.

Q94: Do pro recording engineers all endorse DSD?

Misconception: Since DSD is the most "high-fidelity" format, pro engineers must all be DSD fans.

Truth: Pro engineers and mastering engineers' attitudes are far from unanimously enthusiastic:

Bob Ludwig (Gateway Mastering owner, 11 Grammy wins): said in multiple interviews that 24-bit/96kHz PCM is his delivery format for 80%+ of work
Bernie Grundman (Bernie Grundman Mastering founder): "Our workflow converts to PCM via DAD first, does all the work in PCM, then converts to DSD at the end"
Bob Katz (*Mastering Audio* author): wrote that "DSD's lack of editing convenience is impractical for audio engineers"

Engineers' core reason for not championing DSD is the repeatedly mentioned inability to edit DSD. Their PCM preference isn't "not understanding higher specs" but that PCM offers a better workflow and easier QC.

Q95: Is the "Hi-Res Audio" label an authoritative certification?

Misconception: "Hi-Res Audio" is a rigorously vetted certification; DSD with the label guarantees quality.

Truth: JAS (Japan Audio Society) and CEA (US Consumer Electronics Association) jointly defined "high-resolution audio" in 2014: any recording format better than CD quality.

Meaning:

DSD64 (2.8MHz/1-bit) qualifies
PCM 24-bit/96kHz qualifies
PCM 24-bit/48kHz? Actually not — 48kHz's 24kHz Nyquist matches CD's 22.05kHz range

But the label's bar is extremely low — technical spec compliance only; no recording-quality verification, no source provenance audit, no format authenticity review. So:

A DSD upsampled from 16-bit/44.1kHz can carry the "Hi-Res Audio" mark
MP3-upsampled DSD can too

The "Hi-Res" label is a format-threshold certification, not a quality certification — like "high resolution" doesn't guarantee a good photo.

Q96: Is buying an expensive DSD player the only path to quality upgrade?

Misconception: You must buy a dedicated DSD player to hear DSD — the right path to upgrading.

Truth: Entry HD players (ELAC Discovery, SMSL entry DSD DACs) play DSD512 within a few thousand yuan. High-end units (dCS Bartók, MSB Discrete, CH Precision C1.2) cost hundreds of thousands — but the DSD portion is a small fraction of the cost.

At the same budget:

Upgrade your main monitors → soundstage, clarity, bass layering — 100% visible improvement
Add proper room acoustic treatment → improves early reflections/reverb that 80% of people hear — a "quantum leap"
Buy a decent XLR digital cable (not too expensive) → signal integrity (110Ω balanced per AES/EBU)
Only last, change the DAC → marginal improvement after everything above is optimized

Reversing this order (upgrade the DSD DAC to the highest first, then speakers/acoustics) is classic "consumer priority misalignment."

Q97: Are devices without DSD support outdated?

Misconception: A 2024 DAC without DSD support is technologically behind and should be retired.

Truth: Pro studio monitor DACs (Lynx Hilo, Prism Sound Dream DA-2, RME ADI-2 Pro) had limited early DSD support. Their PCM performance remains industry reference-grade today.

Today, lacking DSD does affect consumer experience (much "audiophile high-res" music ships as DSD), but it's unrelated to "advanced/outdated." A device can:

Have no DSD, yet PCM 24/192 THD+N < -115dB
Have DSD, yet PCM 24/192 THD+N > -90dB

Clearly the former's technical capability far exceeds the latter's. "No DSD support" is more a convenience issue than a technical-level issue.

Q98: Do DSD marketing claims all have technical basis?

Misconception: Vendor claims — "direct stream," "high-precision 1-bit," "no quantization error" — are verified technical facts.

Truth: These are the classic claims worth reading together (among the 130+ misconceptions):

"Direct Stream" — unrelated to path directness; DSD paths are usually more circuitous
"1-bit high precision" — 1-bit is the lowest precision, not the highest
"No quantization error" — quantization error is inherent to digital systems; DSD included
"All-DSD workflow" — internal processing passes through the PCM domain
"Beats high-spec PCM" — unverified on dynamic-range metrics
"DSD is analog flavor" — measurable distortion-pattern changes, not analog flavor

If a product's marketing combines such phrases, be wary. Not all such vendors are frauds — some engineers genuinely believe these statements — but consumers should keep critical thinking.

Chapter 7: Misconceptions About Format Comparison (9 items)

Q99: Is DSD better than PCM in every way?

Misconception: As a "new-generation" format, DSD must beat PCM.

Truth: DSD vs PCM must be compared per parameter and per playback system; no format wins everywhere:

Dimension	DSD64	PCM 24/96	Winner
Audible-band dynamic range	≈120dB (after shaping)	144dB (inherent)	PCM
THD+N (typical)	-95~-105dB	-110~-120dB	PCM
Low-frequency response	No difference	No difference	Tie
True high-frequency signal retention	Limited by noise layer	Unrestricted	PCM
Post-editing convenience	Needs PCM conversion	Native	PCM
Storage efficiency	2.8Mbps uncompressed	~1Mbps FLAC	PCM
Noise shaping for PCM conversion	Noise layer retained	No conversion needed	PCM
Playback device ubiquity	Niche	Universal	PCM
Listening preference on specific DACs	Some users prefer	Some users prefer	Subjective

This isn't "one side crushes the other." On objective measurements and practical ubiquity, PCM has clear advantages. But objective measurement isn't identical to listening evaluation. Some DACs use different analog output designs per mode (switch-resistor networks vs multi-bit R-2R) — these DAC-architecture differences may affect listening more than the theoretical format gap. Differences between DSD and PCM modes in one DAC mix format causes with circuit-design causes — often conflated in comparisons.

Q100: Is DSD the perfect replacement for PCM?

Misconception: DSD is PCM's natural evolution and should fully replace it.

Truth: DSD hasn't and can't fully replace PCM. Technically simple:

Information theory: DSD's 1-bit architecture is less efficient in information entropy than multi-bit PCM. Matching dynamic range requires higher Nyquist bandwidth and more complex noise shaping
Engineering practice: PCM math (FFT, convolution, filtering) is far simpler; DSP chips are highly optimized for PCM
Network effects: all-digital distribution infrastructure (CD, DVD, Blu-ray, streaming, downloads) is PCM-based. Full DSD migration means rebuilding infrastructure — impossible

For the foreseeable future, PCM remains digital audio's "base layer"; DSD appears as a "transport-layer format" between front and back ends — not PCM's replacement.

Q101: Is high-spec PCM (e.g., 24/192) worse than DSD?

Misconception: 24/192 PCM may measure well but doesn't sound as natural as DSD.

Truth: A typical "listening > specs" argument. 24/192's 192kHz rate, 96kHz Nyquist, 144dB dynamic range — every objective metric massively beats DSD64.

"Why it sounds less natural" possibilities:

Upsampled version quality — the PCM file compared may itself be a "fake high-res" upsampled from 44.1kHz to 192kHz
DAC frequency-response differences — output-stage response differs between modes
Low-pass design compromise — some DACs use gentler PCM filters (slow rolloff) limiting high extension, while DSD mode uses different filter design
Matched subjective preference — users may prefer DSD mode's specific distortion character and call it "more natural"

Without strict level matching and blind control, any "24/192 sounds worse than DSD" conclusion is unreliable.

Q102: Is CD format obsolete, DSD the future?

Misconception: CD 16/44.1 should be retired in the 2020s; DSD is the future standard.

Truth: CD-standard 16-bit/44.1kHz PCM remains (and will remain) audio's "golden reference." Streaming platforms' (Tidal, Qobuz, Deezer, Apple Music) base lossless tier is CD quality (FLAC 16/44.1) — about 1.4Mbps.

DSD's "adoption probability" in the mainstream consumer market is near zero because:

File sizes are too large (DSD64 is 2-3× H.264-compressed CD FLAC)
Streaming transport is inefficient (no efficient DSD compression — DST averages only 2:1 vs FLAC's 4:1-6:1)
Hardware support is low (of the world's hundreds of millions of playback devices, DSD-capable units number under ten million)
Production cost is high (reduces studio efficiency)

DSD won't replace PCM in the mainstream, but it has an independent, stable niche in audiophile and high-res playback — especially as PCM→DSD conversion quality and DSD DAC design keep improving. It's not "does it have a future" but that it never needed to "become mainstream" to justify its existence.

Q103: Is analog-to-DSD transfer better than to PCM?

Misconception: Digitizing analog masters (reel, tape) with DSD preserves more of the original than PCM.

Truth: Analog-to-digital is AD conversion; the AD converter's quality (not the output target format) is decisive.

Whether the target is DSD or PCM, the core AD steps are:

Analog → preamp → anti-aliasing filter
ΔΣ modulation at high rates (5.6MHz+) → multi-bit internal representation
Downsampling/re-modulation → target format (PCM or DSD)

At step 2, the signal already exists as multi-bit PCM equivalent to the AD converter's internal precision (usually 24-32-bit float). Whether the final output is DSD or PCM, the internal representation is identical.

What truly matters is the AD converter's analog front end — transformer quality, preamp design, clock precision. None of this depends on output format. "Analog to DSD is better" doesn't hold.

Q104: Is the DSD-vs-PCM debate settled?

Misconception: After 20+ years, consensus exists that DSD or PCM is better.

Truth: The debate has no conclusion — and may never. Not because the formats differ too much, but the opposite: the difference is too small.

When formats are indistinguishable in controlled blind tests, the dispute moves from "objective fact" to "subjective preference." One person prefers DSD, another PCM; both can support their judgment with objective data — under a specific harmonic distortion profile some users prefer DSD, under other conditions PCM. Neither is wrong (nor right).

More sensibly: DSD isn't PCM's enemy but its variant. In the right scenarios (AD converter output, SACD release, rare all-DSD recordings), DSD may hold slight advantages. In most others, PCM is more convenient, cheaper, higher-performance. Debating "which is better" is like debating "train vs plane" — it depends on where you're going.

Q105: Are new formats like MQA worse than DSD?

Misconception: MQA is lossy; DSD is lossless; so DSD is better.

Truth: MQA and DSD have different goals and technical routes:

MQA: encodes high-res audio into lower rates (e.g., CD 44.1kHz) via folding, then unfolds in MQA decoders. Its design goal is streaming efficiency + authentication; core is lossy folding
DSD: 1-bit lossless pulse flow; no compression of the original signal, but files balloon

Comparison across dimensions:

Preserving native recording quality: DSD64 on untouched original DSD material theoretically preserves the full 1-bit original; MQA imposes lossy filtering on CD signals through fold/unfold
Practicality: MQA transports high-res at CD-level bitrates; DSD can't carry equivalent information below 2.8Mbps

Most importantly: DSD and MQA aren't "one replaces the other." Both are nodes in the digital distribution pipeline, not endpoints. The deciding factors should be whether your gear fully supports it and personal storage preferences — not superstition that one format beats another everywhere.

Q106: Is DSD the audiophile's only choice?

Misconception: The audiophile's ultimate pursuit is DSD; high-spec PCM is "just specs."

Truth: In the top audiophile circle, DSD and PCM each have half the audience — and many top systems don't rely on DSD. Influential references:

Steve Guttenberg (The Audiophiliac): his main reference system is PCM-based
Paul McGowan (PS Audio founder): uses DSD as his main reference but doesn't deny PCM's quality
Robert Harley (*The Absolute Sound* editor): wrote "a good preamp/amp design affects sound 10× more than format choice"

DSD isn't the audiophile's only choice. It can be *a* choice, but shouldn't be labeled "the one right path."

Q107: Will all music be released in DSD in the future?

Misconception: As storage costs fall and bandwidth rises, all releases will go DSD.

Truth: Storage and bandwidth are just one obstacle. The bigger one: the entire content-production ecosystem is built on PCM.

If a song is (1) recorded in PCM (2) mixed in PCM (3) mastered in PCM — releasing it as DSD is "gilding the lily" — the quality ceiling is set by the PCM intermediate stages. Conversely, skipping PCM for all-DSD production runs into DSD's editing limits.

DSD-incompatible factors:

The industry's all-PCM infrastructure
Zero DSD support in streaming codecs (AAC/Opus/FLAC aren't DSD)
DSD encryption/watermarking never practicalized for digital distribution
No unified streaming codec standard for DSD

DSD won't replace PCM as the mass streaming format, but in high-res playback and audiophile scenarios, PCM→DSD upsampling and SACD releases maintain an independent, active niche.

Chapter 8: Misconceptions About Technology and Mathematics (8 items)

Q108: Is DSD's mathematics more advanced than PCM's?

Misconception: 1-bit ΔΣ modulation is a more advanced mathematical method than PCM's linear sampling.

Truth: Mathematically:

PCM uses uniform quantization: a signal's amplitude at a time-domain point quantizes to the nearest digital value. Noise is uniform in frequency; level depends on bit count
DSD uses 1-bit ΔΣ modulation: the same time-domain point isn't quantized alone but fed back with history and prediction

Both rest on the Nyquist-Shannon sampling theorem. PCM applies it directly; DSD achieves the goal indirectly via oversampling + noise shaping. Neither is "more advanced" — they're different engineering trade-offs.

ΔΣ modulation's theoretical basis dates to the 19th century — it isn't a modern invention.

Q109: Is noise shaping DSD's secret technique?

Misconception: Noise shaping is DSD's exclusive core technology; PCM doesn't have it.

Truth: Noise shaping is common in PCM too. In fact, all modern DAC chips (ESS, AKM, Burr-Brown, Cirrus Logic) use noise shaping in their PCM-mode ΔΣ modulators — converting 24-bit PCM to lower-bit internal representations (5-7 bit) with high-frequency noise shaping.

In the PCM domain, noise shaping applies to:

Dithering: adding noise-shaped dither during bit-depth reduction (24→16-bit) to lower audible noise
DAC internal processing: multi-bit PCM enters the DAC, the modulator converts to lower-bit higher-rate pulses with noise shaping preserving in-band SNR
Post-quantization processing: some DAWs (iZotope RX) apply noise shaping to quantized audio

Noise shaping isn't DSD's "secret technique" — it's a DSP tool both formats can (and do) use.

Q110: Does DSD's 1-bit stream contain all audio information?

Misconception: DSD's 0s and 1s flow at 2.8M pulses/sec, each derived from the continuous waveform — all information embedded in the stream.

Truth: The 1-bit stream isn't a simple 0-1 sequence at output. To reconstruct audio from it, the decoder must:

Convert the pulse-density signal to analog voltage (via low-pass)
Filter out-of-band noise through the analog low-pass
Recover the audible band's amplitude and phase

During this, the raw 1-bit info is fully preserved, but what signal recovery uses isn't "encoded audio information" — it's "the modulator's output." Viewing the DSD stream on a digital scope shows an almost random 0/1 sequence — audio information is implicitly embedded in pulse density, not explicitly encoded.

This means DSD stream analysis must be based on post-low-pass results, not the stream itself — fundamentally different from PCM's explicit quantized-value encoding.

Q111: Does DSD need no complex digital signal processing?

Misconception: DSD's 1-bit stream goes straight to the DAC and sounds — no complex filtering like PCM.

Truth: DSD's DSP needs are more complex and specific:

Modulator level: the ΔΣ modulator itself is a complex feedback system — noise transfer function design, stability analysis, high-order loops (up to 7th)
Clock recovery: DSD's timing demands are harsher than PCM's — a pulse's width directly determines energy
Low-pass design: as noted, DSD needs especially steep analog low-passes to suppress out-of-band noise — harder than PCM's reconstruction filter

DSD isn't "free of complex DSP" — it just moves the burden from the computer into the DAC.

Q112: Is DSD less sensitive to clock jitter than PCM?

Misconception: DSD is 1-bit; each pulse's width determines energy, so jitter tolerance is higher.

Truth: DSD's jitter sensitivity equals or exceeds multi-bit PCM's:

In DSD, signal energy encodes directly in pulse width/position. If a rising edge shifts 200 picoseconds (10⁻¹²s) from jitter, the ideal pulse waveform shows an extra positive/negative voltage pulse — directly audible-band noise.

Theoretical analysis: DSD64's effective rate is 2.8224MHz; 200ps jitter yields ~-130dBFS noise in the audible band (below 20kHz) — under DSD64's 120dB dynamic range, but for high-spec DSD (DSD256/512), jitter-induced noise rises by a dozen dB, degrading overall noise performance.

Phase-noise requirements for DSD and PCM have no essential difference — a good clock is a good clock, format-independent. On DACs with weak clock-recovery design, jitter's impact on DSD playback can be even more pronounced.

Q113: Is DSD the ultimate "pure digital," free of analog flaws?

Misconception: Since DSD is all-digital from recording to transport, it's free of analog-circuit distortion and nonlinearity.

Truth: DSD's D/A output inevitably needs analog circuitry to drive speakers. Its quality directly affects DSD listening.

The DSD DAC's low-pass uses:

Op-amps — introduce harmonic distortion and noise
Capacitors/inductors — phase shift and nonlinearity
Relays/switches — analog attenuators in DSD volume control
Signal cables — poor shielding picks up external noise

DSD uses no *less* analog circuitry than PCM — similar types (with higher-order, stricter low-pass). And because DSD's out-of-band noise energy is extreme, analog design margins must be larger, with tougher layout and shielding demands. Not "digital straight to the end" but "the unavoidable analog path, narrower and harder."

Q114: Can DSD perfectly record waveforms?

Misconception: The 1-bit stream reproduces the original analog waveform "as-is."

Truth: No digital format (regardless of bit depth) "perfectly" records analog waveforms — digital is discrete, analog is continuous; conversion error always exists.

Mathematically, DSD and PCM have no essential difference: within a finite bandwidth (DSD64's audible band ≤20kHz), approximate the original at sufficient resolution. DSD's approximation is pulse-density modulation (PDM); PCM's is pulse-code modulation. Different error characteristics per dimension, but neither is "perfect."

In fact, DSD64's harmonic distortion across orders sits around -105dBFS (DpdoEngine measured), while the original analog signal's distortion may be

be as low as -120dBFS — the conversion itself loses roughly 10dB of fidelity.

Q115: Is DSD's filtering simpler than PCM's?

Misconception: DSD needs no digital anti-aliasing filter (PCM does), so the path is shorter.

Truth: PCM needs anti-aliasing filters (AAF) at AD, and reconstruction filters at DA. DSD equally needs an AAF at AD (otherwise the 1-bit modulator produces very high IMD), and at DA needs an even steeper low-pass to suppress out-of-band noise.

Total filter count:

Stage	PCM	DSD
Pre-AD anti-aliasing	1	1
Post-AD reconstruction	1 (digital)	1 (digital, downsampling)
Post-edit resampling	Several (workflow-dependent)	Several (PCM back to DSD)
Pre-DA	1 (ΔΣ modulator PCM→pulse)	0 or 1
Post-DA	1 (sinc reconstruction)	1 (multi-order low-pass)

Overall, DSD's filter count is no fewer than PCM's — in some scenarios more/complex.

Chapter 9: Misconceptions About History and Development (5 items)

Q116: Is DSD a 21st-century technology?

Misconception: DSD is a new invention of this century.

Truth: DSD's commercial standard was completed in 1996 (SACD spec); mass productization in 1999 (first SACD players). DSD is essentially 20th-century technology.

Key paper timeline:

1962: Inose, Yasuda, Murakami publish the foundational ΔΣ modulation paper ("A Telemetering System by Code Modulation — ΔΣ Modulation")
1987: Philips files the 1-bit D/A converter patent
1995: Sony and Philips jointly set the SACD/DSD spec
1996: SACD standard released
1999: first SACD players ship

So when people debate DSD in 2024, its technical foundation comes from 62-year-old theory and a 28-year-old standard. Not new technology — just niche tech that never truly entered the mainstream.

Q117: Has DSD technology been developing rapidly?

Misconception: Like CPUs/GPUs, DSD's specs (bitrate, noise-shaping order) keep rapidly upgrading.

Truth: DSD's core 1-bit ΔΣ modulation hasn't fundamentally changed since the late 1990s. Changes are mainly:

Processor performance — modulators now run more complex high-order algorithms
DAC implementation — ASIC to FPGA migration allows more flexible architectures
Order increases — from initial 3-5th order to today's 7-9th
Asynchronous sampling — DSP replacing analog PLL for jitter attenuation

But DSD's core bottleneck — the information limit of 1-bit — has no physical/mathematical breakthrough. Effective dynamic range gains come mainly from higher oversampling rates (DSD128/256/512) — not a product of "DSD technology development" but the bandwagon result of pushing DACs to higher clocks. Higher-clock DSD demands more clock precision — the same technical trend as PCM's "clock advancement."

DSD's "roadmap" isn't built on a long-term breakthrough R&D platform; its current state approaches a plateaued technology.

Q118: Is SACD DSD's only carrier?

Misconception: DSD = SACD and SACD = DSD; they're the same thing.

Truth: SACD (Super Audio CD) is one physical carrier of DSD, not its only form. DSD exists as:

DSD on SACD discs: CD-sized discs storing DSD stereo and multichannel tracks
DSF files: DSD Stream File — for computer playback
DFF files: DSD Interchange File Format — similar but different metadata model
ISO images: digital SACD copies for archiving/playback
DoP streams: DSD over PCM transport
Embedded DSD: some recorders (Korg MR-2000) output DSD files directly

After the 2000s, with SACD players exiting the market, DSD's biggest distribution path is downloaded DSF/DFF files, not discs. Their provenance may be native DSD or PCM upsampling — exactly like discs.

Q119: Was DSD born to satisfy musicians' artistic pursuits?

Misconception: Sony developed DSD/SACD in response to musicians' demands for better sound.

Truth: DSD/SACD was born of commercial crisis, not artistic demand. In the late 1990s, CD sales declined after physical-format saturation; labels needed a new "carrier upgrade" to spur purchases. The SACD vs DVD-Audio standard war between Sony and Philips was likewise a fight over the next-generation disc market.

If DSD truly served artists' pursuit, it would:

Prioritize studio workflow convenience (rather than developing editing tools only later)
Incorporate engineers' feedback (rather than being called completely unusable)
Ship AD/DA solutions early (rather than releasing DSD Direct only in 2004)

DSD/SACD is essentially a commercially-driven market push; its "higher sound quality" is a marketing lever for consumers, not the result of artist-driven demand.

Q120: Is DSD's history a sound-quality evolution story?

Misconception: DSD witnessed continuous quality evolution from CD to SACD to DSD downloads.

Truth: DSD's history is a mixed work of technical compromise + commercial gamesmanship + format standards, not a simple linear evolution.

Key timeline:

1996: DSD spec completed — targeting CD replacement with the dual hook of higher quality + copy protection
1999: first SACD devices — tepid market response
2003-2005: DVD-Audio vs SACD format war
2007: DVD-Audio fails; SACD keeps a tiny but independent niche
Post-2010: DSD downloads rise — starting from audiophile niche needs
2018-2024: DSD still under 1% of global audio distribution

None of these 15 years' key events was a "sound-quality breakthrough" — all were commercial gamesmanship and market dynamics.

Chapter 10: Misconceptions About Consumer Perception (5 items)

Q121: Is buying DSD an investment in sound quality?

Misconception: Paying more for DSD files is investing in better sound.

Truth: DSD's value isn't simply "better sound" — it depends on the whole playback chain's matching. Before upgrading formats, ask:

What's my current library source? If mostly streaming (Spotify/Apple Music AAC 256kbps), AAC → DSD is a genuinely audible upgrade
What's my playback system? With obvious hardware bottlenecks (integrated sound cards, SBC/AAC Bluetooth, entry USB DACs), the format's theoretical advantage drowns in chain distortion
What's my listening environment? Open offices/buses/cafés run 40-60dB(A) background noise — at that level CD vs DSD differences are fully masked

These aren't reasons to deny DSD but coordinates for rational investment — put resources into the chain's bottleneck first, format second. When the first three links (acoustics → speakers → DAC) are in place, DSD/PCM→DSD upsampling can be the final "marginal improvement." Marginal isn't "zero" — in highly optimized systems, that last 5% is often the "final touch" audiophiles chase.

Q122: Are forum praises of DSD objective and truthful?

Misconception: Those 3,000-word "DSD listening comparison" reports are objective evaluations worth referencing.

Truth: Audiophile forum listening reports face systematic biases:

Sunk cost fallacy: someone who spent ¥30k on a DAC + DSD files is more inclined to believe "it does sound better" to justify the expense
Social desirability bias: in a community, endorsing the mainstream preference (DSD good) wins approval more easily than challenging it
AB comparisons without blind testing: 90%+ of "listening reports" are written under non-blind conditions
Memory error: human sound memory is extremely short (4-12 seconds); A/B switch latency alone can fabricate "differences"
Cultural status signaling: "high-end DSD system owner" status itself biases evaluation

The most reliable approach: read comparisons with blind-test protocols and quantitative data. Purely textual "listening reports" are subjective recounts that can't objectively reflect DSD's real gap vs PCM.

Q123: Can ordinary people easily become DSD appreciation experts?

Misconception: Listen to enough DSD and you'll eventually distinguish it from PCM at one listen, like "golden ears."

Truth: More DSD listening trains sensitivity to specific timbres — but doesn't guarantee reliable ABX discrimination of DSD vs PCM.

In a 2009 limited study, subjects listened 4 hours daily for a week on a DSD + hi-fi system, then took A/B/X tests. Result: even after training, subjects were more confident but accuracy didn't exceed the control group — the overconfidence effect.

But note: ABX design has inherent limits — switch latency flattens auditory-memory transient differences; test material may not be spectrally representative; DAC output-stage differences between modes may be balanced away by design. "Can't reliably distinguish in ABX" ≠ "listening experience is identical" — it only means the difference is insufficient to form a stable identification signal under standardized conditions. For many audiophiles, DSD's "listening difference" doesn't need ABX proof — what matters is they like the difference and know where it comes from.

Q124: Is pursuing DSD the mandatory path for audiophile advancement?

Misconception: Every audiophile must eventually reach DSD, or they're "still stuck at CD level."

Truth: The "advanced stages" of audiophilia should be:

System design (room acoustics, matching principles, listening habits)
Recording research (different editions, mastering, original media characteristics)
System tuning (power, damping, placement, noise reduction, listening position)
Format and conversion experiments (exploring format characteristics and preferences)

DSD can be an important node on the path. "Focusing on DSD" isn't itself a problem — if the audiophile understands DSD's technical boundaries and costs, and genuinely prefers its sound on their system, that's an informed preference choice, not "not yet reaching the advanced stage." The problem isn't DSD itself but the attitude of deifying it as "the one right path."

Q125: Are DSD critics just "can't afford good gear"?

Misconception: DSD critics are people who can't afford high-end DSD DACs and never experienced real DSD playback.

Truth: This is ad hominem rebuttal — attacking the speaker while ignoring evidence.

The most professional, sharpest DSD criticism comes not from "audiophiles who can't afford it" but from:

Pro recording engineers (Bob Katz, Bob Ludwig, Bernie Grundman)
Signal-processing scientists (Xiph.Org's Christopher Montgomery)
Hardware design engineers (DAC/ADC chip designers like Bruno Putzeys)
University audio research professors and PhDs

These people can not only "afford" top gear — their audio systems *are* self-designed reference systems. Their reason for seeing DSD's limits: on measurable technology, it's constrained by the 1-bit foundation.

Chapter 11: Misconceptions About Special Phenomena (5 items)

Q126: Is audible noise in DSD files normal?

Misconception: DSD playback with a faint hiss is DSD's "signature flavor."

Truth: DSD's 1-bit ΔΣ modulation produces an audible noise layer in silence — leakage from the DA low-pass's limited rejection. At DSD64, filter cutoff is ~50-70kHz, but:

Filters have finite high-frequency attenuation; on poorly designed DACs, out-of-band noise above 50kHz can "leak" into the audible band's bottom
High-frequency activity (idle tones) during silence can couple with the output stage, showing measurable voltage fluctuation

Audible "hiss" is usually not DSD's "inherent trait" — it's the result of inadequate DSD low-pass and output-stage design. Better filters and power decoupling suppress it significantly.

In DpdoEngine tests (early v6.50-6.55 data), optimizing low-pass parameters lowered DSD64 silence noise from -90dBV to -115dBV (across two output configs). v6.60's modulator update further improved idle-tone behavior, with a better silence noise floor under the same config.

Q127: Does a "blacker" background after upsampling to DSD mean better quality?

Misconception: Upsampling a CD to DSD with HQPlayer makes the background "blacker" (hiss gone) — quality improved.

Truth: The "blacker background" phenomenon's math:

When CD (16/44.1) upsamples to DSD64, the upsampler inserts interpolated samples into the 44.1kHz signal, then ΔΣ modulates to pulses — this does several things:

Quantization noise redistribution: the CD's quantization noise (-96dBFS uniform white) is reshaped by the upsampler's noise shaping — far more at high frequencies, on average ~12-15dB lower in the audible band (below 20kHz)
Removal of high-rate quantization noise: CD's high-frequency quantization noise near 44.1kHz is removed by interpolation and filtering
Relative low-band noise reduction: with noise "pushed" to the ultrasonic band, measured in-band RMS noise genuinely decreases

But "blacker background" ≠ "more faithful" — the filtering and noise shaping during upsampling also change the original signal's time-domain response (e.g., impulse response). Upsampler design quality determines the outcome — a poorly designed upsampler may introduce more time-domain distortion than benefit.

"Blacker background" just means the upsampler changed the noise distribution — the format itself didn't "restore detail." On some upsamplers, a quieter background may come at the cost of the original music's transient response — "quieter noise," not "better reproduction."

Q128: Is upsampling to DSD with HQPlayer the ultimate play?

Misconception: HQPlayer + DSD512 = digital audio's ultimate playback.

Truth: HQPlayer is indeed an excellent high-quality player; its upsampling algorithms (sinc filters, modulator parameter tuning) beat most free software and built-in DAC upsampling. But "HQPlayer to DSD512 is ultimate playback" lacks supporting evidence:

DSD512's out-of-band noise energy is extreme (~-40dBFS above 50kHz), demanding enormous DAC analog performance — most commercial DACs test worse THD+N at high-rate DSD than DSD64
DSD512 files are huge with negligible listening difference: as noted, blind tests can't reliably distinguish DSD64 vs DSD512 — the better the upsampler, the lower the marginal benefit of "more upsampling"
Upsampling changes the sound: HQPlayer doesn't leave audio data untouched — its sinc filters, noise-masking and modulator choices form different processing chains altering time-domain character. Some like the change, but it's not "more faithful"

Compared with blind-testable DSD upsampling gains, bigger quality improvements come from fixing the source — better recording editions, better acoustics.

HQPlayer is a superb DSP tool, but deifying it as "ultimate" ignores a basic fact: in the signal chain, the analog circuitry after the DAC (amps and speakers) affects overall system quality far more than the format conversion before the DAC.

Q129: Is DSD's lower level a flaw?

Misconception: DSD playback levels are lower than PCM, so DSD's "dynamic range is wasted."

Truth: DSD/PCM level differences usually come from different mastering standards, not the format:

CD mastering standard: peaks usually -0.5dBFS to -1dBFS (near max 0dBFS)
DSD mastering standard: peaks usually -3dBFS to -6dBFS, leaving more headroom

This isn't a flaw nor inevitable — it's a differentiated habit by mastering engineers for DSD, because:

DSD modulator stability degrades near 0dBFS, potentially showing visible distortion at peaks
More headroom accommodates transient peaks without clipping
DSD's low noise layer means headroom doesn't make background noise audible

If users find the "level low," equal-gain compensation at playback works — just note DSD-domain volume adjustment may introduce the pulse-modulation noise mentioned earlier. Level difference is a mastering-habit issue, not a format issue.

Q130: Does spectrum content above 22kHz mean a good recording?

Misconception: A spectrogram showing extension above 22kHz (to 40kHz+) means high-quality, "high-spec" recording.

Truth: Human frequency perception tops out at 20kHz (most adults roll off significantly by 15-17kHz). Content above 22kHz may come from:

DSD's noise layer: noise shaping pushes quantization noise above 20kHz — unrelated to music; not "better recording," just noise
Upsampling interpolation residue: PCM-upsampled DSD or high-spec PCM may show computational interpolation signals above 22kHz — not originally in the recording
Rare limiter harmonics: in very rare cases, instrument high-frequency overtones extend to 40kHz — but inaudible
Mic/amp/ADC ultrasonic noise: this also isn't "higher recording quality"

High-frequency extension on a spectrogram isn't itself a recording-quality metric. Real criteria:

Source file quality (analog/digital? master generation?)
Recording technique (mic placement, room acoustics)
Mastering prudence (over-compression? processing artifacts?)

Extension above 22kHz — especially from DSD noise layers — at least lets consumers judge format authenticity, but never equals "better recording."

Summary and Recommendations

130 misconceptions laid out may seem dizzying. But distilled, the core reduces to five questions:

Q1: What level is your playback system?

→ Phone headphones/computer speakers/Bluetooth? The chain's losses — system noise floor (typically >30dBA), DAC THD+N (-80~-90dB), Bluetooth codecs (AAC/SBC/LDAC) — already exceed the theoretical DSD vs FLAC CD gap. DSD makes no audible difference here; this conclusion depends not on listener quality but the chain's physical bottleneck.

Q2: Do you have trustworthy DSD source files?

→ Files "picked up" from forums/cloud drives are 99% upsampled, no better than CD. Even verified purchased albums are rarely native DSD. Ensure spectral analysis and source provenance tracing.

Q3: Can you reliably distinguish in blind listening?

→ If you can't maintain >75% accuracy in ABX double-blind tests (50+ trials), DSD's "audible improvement" doesn't hold. No "golden ears" needed — anyone can try.

Q4: Where can your chain improve most?

→ Speakers/headphones → room acoustics → amp → DAC → power → digital cables → format. DSD (and PCM→DSD upsampling) is the last link — but that last 5% is sometimes the "final touch" in certain systems.

Q5: What are you willing to pay for DSD?

→ If the four answers are "system adequate"/"files trustworthy"/"distinguishable"/"other links topped out" — then DSD is worth pursuing, with rational expectations of marginal gains.

About DpdoEngine

This document is based on measured data from DpdoEngine v6.60 development. DpdoEngine is one of the few complete toolchains on consumer hardware supporting both PCM→DSD and DSD→PCM bidirectional conversion, integrating FIR upsampling, ΔΣ modulation, noise shaping, Kakeya polynomial compression, DST codec, SACD ISO authoring and THD+N analysis. Digital filters support AVX-512/AVX2/SSE4.1/NEON multi-platform SIMD acceleration, with FIR taps up to 131072 (--hb mode), 65536 default. We accumulated extensive comparison test data during development; all data in this article is reproducible.

Our attitude toward DSD: it's an interesting format with reasons to exist (especially as SACD's transport format and part of specific recording workflows), but it shouldn't be deified. Most of the marketing narrative built around DSD fails technical scrutiny. We hope this document helps listeners find rational coordinates in the DSD fog.

130 Questions About DSD: The Most Common Misconceptions About Direct Stream Digital