How do microphones, studio conversion and mastering limit what a DAC can reproduce?

Overview. Watts describes the recording chain (microphones and their placement, the ADC, mixing, mastering) as part of what a listener hears but cannot know: in a podcast published in October 2026 (recording date not stated) he said a listener was not present at the recording, so does not know the acoustic, microphones, cables, tape or mastering processing Part-Time Audiophile 6:40. The link he describes most specifically is the ADC, and the limit he calls irreparable by a DAC filter is aliasing Head-Fi, Oct 2017Head-Fi, Oct 2018. His most concrete listening evidence comes from his own test recordings from 2022 onward Head-Fi, Mar 2022.

2014 to 2016: ADC noise and aliasing, before his own recordings

In June 2014, in the Hugo thread, he said "As a rule of thumb, the perfect spot is the original recording", and that some high-resolution files are only upsampled and "sound awful". In the same post he said a benefit from 44.1 kHz "also assumes the ADC uses correct anti-aliasing filters, which in most cases they don't", and that "as consumers, we never really know what we are getting" Head-Fi, Jun 2014. The same day he said audio ADCs have noise shapers that add high-frequency noise that "is not present at the microphones"; once that noise reaches the DAC's analog stages it causes noise floor modulation, which he hears as hardness or brightness, and he named DSD64 as the worst case Head-Fi, Jun 2014.

In February 2015, again in the Hugo thread, he said "ADC chips do have significant aliasing, and normally the chip designer says its OK as the errors are at say 21 kHz", and that this is audible because it changes time domain linearity Head-Fi, Feb 2015. In March 2015 he said that if a 96 kHz master was converted to 44.1 kHz with a poor sample rate converter, "no amount of increased taps will solve the SRC distortion and noise", so the answer "depends upon the recording" Head-Fi, Mar 2015.

He described his DACs' HF filters as aimed at ADC noise. In December 2015, in the Dave thread, he said Dave's 60 kHz filter was meant to "bandwidth limit high sample rate recordings - DXD and 384k have huge amounts of noise shaper noise from the ADC" Head-Fi, Dec 2015. In March 2017, in the Hugo 2 thread, describing an HF filter he gave as second order with -3dB at 40 kHz (the green trace in the plot under discussion), he said it makes "a very big difference with 96 and 192 - its easy to hear" but "very much smaller with 44.1", because "this noise is artificial, as it is ADC noise shaper noise" Head-Fi, Mar 2017.

In February 2016, in the Dave thread, he said the ADC he was designing would solve "huge noise floor modulation, poor anti-aliasing filters, and poor noise shaper performance" in conventional ADCs Head-Fi, Feb 2016. The next day he gave its project code word as Davina and said normal ADC decimation filters "are half band, so offer worst case only -6dB rejection", adding that with -140 dB decimation filters he could "still hear the effects of aliasing" Head-Fi, Feb 2016. In November 2015 he had said "we have no control over the recording chain, and the loudspeakers", which he gave as an attraction of building a professional ADC Head-Fi, Nov 2015.

2016 to 2017: recording quality over format

In March 2016, in the Dave thread, he said "Probably the best recordings I have is 192/24 - and generally, all things being equal, higher SR is preferable - but not by much", and that the professional ADCs then available do not have zero aliasing: "its as bad as -6dB" Head-Fi, Mar 2016. In May 2016 he said "The recording quality with PCM I think is more important than the actual format", and that he has redbook recordings that sound better than DXD ones Head-Fi, May 2016.

In January 2017, in the Dave thread, he said "my feeling is that the standard is much less important than the recording method". He said Decca recordings from 1960 to 1967 "do things that modern 192/24 kHz recordings are incapable of doing", notably the "speed and impact of real instruments", and that "simple microphone and short signal paths with sound optimized custom built mixers and amps are more important than 44.1 or 192" Head-Fi, Jan 2017. Later that month, about the M scaler, he said "the format is not important as the original recording is way more important" Head-Fi, Jan 2017.

On why those older recordings sound as they do, he hedged. In January 2017 he wrote "I suspect it is Decca's tape recorder modifications, and Mercury using 35mm tape that is playing a key role here - its not just mic technique", then floated an idea about tape bias timing errors and added "Just a thought" Head-Fi, Jan 2017. In August 2017, in the Hugo 2 thread, he said "some of the magic is down to simplicity of both mic technique and simple tuned electronics, but it's not all down to that", and blamed modern ADCs: "the chip designers have forgotten their basic digital sampling theory in that they allow unacceptable levels of aliasing as they all use cheap half-band filters" Head-Fi, Aug 2017. In the same post he said of the Decca box sets not being re-mastered that "this I think is a huge benefit" and said the current Mercury box sets had been re-mastered while "the older CD versions sound much better" Head-Fi, Aug 2017.

On microphones directly: in September 2016 he said, on microphone choice for his planned recordings, "A bright sounding mic will not help and will foreshorten depth perception" Head-Fi, Sep 2016. In January 2017 he said that comparing a show system to a live ukulele has a "flaw" in that "the microphone and ADC is also part of the chain" Head-Fi, Jan 2017. In February 2017, on whether ultrasonics matter, he said "my answer is conjecture, and I know nothing for sure" and that he chose microphones flat to 50 kHz so he could test it Head-Fi, Feb 2017.

2017 to 2019: what the DAC cannot repair

In October 2017 (Watts Up thread) he said the only aspect that potentially messes up the WTA filter is aliasing, because modern ADCs "pretty much all use poor quality half band filters (-6dB at FS/2)"; "no filter can remove aliasing", and the damage is irreparable. He added that "every single track sounds better using a BluDave" in his roughly 5TB library, so the filter still helps aliased recordings Head-Fi, Oct 2017.

In October 2018, in the M scaler thread, he said the M scaler fixes the sampling-in-time problem, but that "modern ADC's are poor, using half band decimation filters (only-6dB at FS/2)", so "the signal will have timing issues that the M scaler can't fix - once aliasing happens it is permanently damaged" Head-Fi, Oct 2018. He said early digital recorders were better because they used analog anti-aliasing filters, and that he "generally" prefers modern quality recordings, which give "improved soundstage, instrument separation and focus" while older ones can have more speed and timbre variation. On whether that older-recording advantage comes from simpler gear or from aliasing, he said "I suspect it may be bit of both" Head-Fi, Oct 2018.

In July 2018, in the Watts Up thread, he described the recordings in use as, in his view, "seriously flawed", with ADCs that have "huge aliasing issues (creating transient timing errors), large amounts of measurable noise floor modulation, and measurable small signal amplitude nonlinearity" Head-Fi, Jul 2018. In June 2019 he said "the vast majority of recordings are via N bit delta-sigma ADC modulators" and that even most DSD recordings are converted from PCM for editing, "with a DXD file used as the master" Head-Fi, Jun 2019. The same month he said modern ADCs "use half band filters because they are cheap", with the level at 24 kHz only -6dB down, so a 96 kHz file down-converted to 48 kHz picks up aliasing that changes transient timing Head-Fi, Jun 2019.

Mastering and processing

In February 2020, in a reply about volume and DSP, he said "if the mastering is done with DXD then you are always better to use that file - always try to get the mastered file whatever the format" Head-Fi, Feb 2020. In November 2024 (Mojo 2 thread) he said "my advice is always send my DACs the original recording in the format that it was mastered in" Head-Fi, Nov 2024. In a September 2022 CanJam talk, when a host noted that labels keep remastering, he said "Remastering is better. Sometimes it's not. Let's go back to the original" (the sense of the first sentence is unclear, so check the recording) The Source AV 32:45.

In August 2023 he said it was not quite right that he thinks early Decca sound better: "one quality sticks out that is not present on modern good recordings", namely "speed, sharpness or impact", which he said he knew is "an attribute to aliasing and transient timing reconstruction". He also allowed that "it is nothing to do with that, but an EQ or HF distortion issue", and closed with "we can't be certain about this, and I suspect that both factors are at play with the better aliasing being the dominant part" Head-Fi, Aug 2023. He added "without being there, and knowing the whole chain, it's difficult to state anything with any kind of certainty" Head-Fi, Aug 2023. In a follow-up he said the fast-edged Decca sound comes from pre-Dolby analog tape (before 1968) and "I hated (and still do) the Decca digital recordings" Head-Fi, Aug 2023.

In February 2023 (Dave thread) he said that with master tapes before 1968 "instruments sound real", and "The minute that you put the master tape through an ADC and then replay back with a conventional DAC, you lose those special qualities"; in the same sentence he placed "the unique problem with digital" in the DAC's reconstruction of transient timing Head-Fi, Feb 2023. In the same month he said that his son, a sound recording student, told him "classical recordings always use mild compression", and that he would not use it himself ("I certainly won't use compression"). He also said the roughly 100 kHz bias on reel-to-reel recording heads "has the effect of distorting the timing of transients", and that he had often thought "perhaps that is why people like the sound of reel to reel" Head-Fi, Feb 2023.

On mixing, in the March 2022 recording post he said "Most recordings are multi mics, with huge mixing consoles" Head-Fi, Mar 2022. In a HiFi Buys interview published in October 2026 (recording date not stated; its transcript places it before May 2026) he described a studio multitrack in which one instrument on one microphone has immense clarity, but "a minute you start mixing everything together, it just all falls apart" HiFi Buys 1:07:10.

2022 to 2023: his own Blumlein test recordings

In March 2022, in the Watts Up thread, Watts described recording an orchestra practice session with a stereo ribbon microphone in the Blumlein arrangement into a Focusrite Scarlett 18i8, monitored through Hugo 2 and DCA Stealth headphones. He reported that "digital monitoring sounded very close to the actual sound" and that left-to-right placement showed "no detectable difference in angle" Head-Fi, Mar 2022. Depth was the main shortfall: the lead cello, about 5 meters from the microphone, "dropped by about 50%, to say 2.5m impression of depth" on headphones; back home, Hugo 2 with the Stealth gave around 50 percent, and what the post calls "x Scaler/Dave" gave around 66 percent, "perhaps better but no more than 75%" Head-Fi, Mar 2022. (In April 2022 he named an "xM scaler" as one of his current projects Head-Fi, Apr 2022.)

He also reported that "All the takes had a softening to transients" and that playback was warmer than the live sound, in a church he described as "quite bright" Head-Fi, Mar 2022. His own account also gives microphone position a role: later takes had "the rear lobe of the mic pointing more to the ceiling, thus suppressing the perception of depth" Head-Fi, Mar 2022. His attribution: "Now the above deficiencies were entirely expected and I am convinced due to the ADC", naming aliasing, which "directly affects the perception of the starting and stopping of transients", and noise floor modulation, "something for which modern ADCs have large amounts of" Head-Fi, Mar 2022. He concluded that Blumlein recording "doesn't have any fundamental flaws" Head-Fi, Mar 2022.

Follow-ups repeat this. In April 2022 he said the deficiencies "are in the ball park for my expectations of the sonic losses of the Focusrite ADC" Head-Fi, Apr 2022. In May 2022 he said "when I did my recent recording, it was clear that conventional ADCs soften the sound", and in the same post said that minimum phase, MQA, apodizing and NOS filters "seriously degrade transient timing, which radically softens the sound - vastly more than the problems of ADCs" (listening tests on his M scalers) Head-Fi, May 2022. In July 2022 he said the recording played back "closer to the original sound than I expected" and that the deficiencies were "issues I expect from the ADC - that is reduced soundstage depth and soft transients with a warmer balance" Head-Fi, Jul 2022.

In an interview published in August 2022 (recording date not stated) he said something recorded four meters away "sounds two metres away when you're listening to it on headphones" and that "I'm pretty sure that's down to the A to D converter"; he added that the softness of transients is also down to the converter, "because I know what the effects of decimation is doing" (check the recording, and note that the written March 2022 post gave about 5 meters) AVstore 47:20AVstore 48:06. In a September 2022 talk he said "The small signal accuracy isn't very good, and that's the reason why depth is being suppressed, and the aliasing is awful on an A to D converter" The Source AV 20:30.

In January 2023, in the Mojo 2 thread, he said of these recordings that "digital is too soft and warm, with degraded timbre variation, and has much degraded soundstage depth", while lateral imagery was "spot on perfect" Head-Fi, Jan 2023. Two days later he said "ADCs are severely limiting performance as they suffer from noise floor modulation, poor (in my terms) small signal accuracy, and hideous amounts of aliasing", while agreeing it is "a whole chain of mic (including placement), cabling, ADC, recording software, lap-top and powering, DAC/amp and headphones"; he said the lateral imagery needed no improvement Head-Fi, Jan 2023. In a February 2023 Q&A he said of the ADC "It's got a softening sound to it for various reasons" and that his ribbon microphones, depending on placement, "can sound a little bit warm" Soundnews 4:25.

2025 to 2026: how much is the ADC, and his own ADC

In an ecoustics podcast recorded in October 2025 (published March 2026) he said "the biggest problem is on the DAC side" and "my estimate is around 80 percent of the sound quality losses is on the DAC side and twenty percent or 25 percent" on the ADC side, based on "my own recordings, not using my A D C, but using commercial available A D C's" (readers should check the recording for the figures). He said this after agreeing ("Oh, absolutely") with the host that the ADC stage is very important ecoustics 20:11. In the same interview he said he was developing a pro audio ADC and "we'll be doing recordings fairly shortly" with it ecoustics 18:59, and that recording techniques, "be it real to real or digital", do not capture sharp sounds properly, so playback sounds softer than live ecoustics 1:15:45.

In a Soundnews conversation recorded in autumn 2026 and published on 17 October 2026, he said "we're now doing 768 kHz recordings with the Pulse Array ADC", to be decimated to 48 kHz and back so the losses can be compared article, Oct 2026. He said recording engineers spend much of their effort "trying to compensate for the losses that the ADC is creating" article, Oct 2026.

In a podcast published in October 2026 (recording date not stated), looking back at the conventional-ADC recordings, he said soundstage depth was attenuated and the digital recording was "very much warmer than the original recording", which "was partly down to the characteristics of the ribbon microphone. But it wasn't the entire thing" Distorsiuni Interioare 11:17. He called the missing speed, impact and power "a reproduction side, a DAC side problem, and it's also very much an A to D converter problem", and said two test recordings had been made with the pulse array ADC in Quartet Distorsiuni Interioare 11:57. The surprise, he said, was the pitch of female vocalists: "It's so much better than comparing it to commercial Pro ADCs", which he attributed to aliasing in standard pro ADCs Distorsiuni Interioare 12:51Distorsiuni Interioare 13:35. Later in the same podcast he said engineers have "an ADC that's too bright because they all are too bright, so they have to tame it by using EQ" Distorsiuni Interioare 1:31:20.

In another podcast published in October 2026 (recording date not stated) he repeated the 2022 finding, "The depth is truncated by about 50 percent", and said it is one of the reasons he has been working on the pulse array analog-to-digital converter Part-Time Audiophile 14:32.

Not knowing what is on the recording

At CanJam RMAF in October 2017 (video published December 2017) he said of a purchased recording that "You don't know what the hell you're listening to", and that when a change to his product makes the variability bigger, "you know that you've actually improved the fundamental transparency of the system" RMAF 7:43. At CanJam NYC in late February 2023 (video published March 2023) he said "You've no idea what the original sound was" and that more variation in timbre, space and instrument separation means "you're getting closer to the original musical truth" The Headphone Show 35:55. In July 2024 he listed what a listener cannot know about a recording: "microphones sound very different, the microphone cables sound different, the mixing desk sounds different", along with the EQ, the compression and the reverb HIFICLUB 16:20. In a podcast published in October 2026 (recording date not stated) he said "all these variables mean that we really don't know" the provenance of recordings Part-Time Audiophile 6:40.

What the items do not show. Watts gives no measured split between the effects of the microphone and its placement and those of the ADC; his 2022 post names both Head-Fi, Mar 2022, and in a podcast published in 2026 he said the warmth was partly the ribbon microphone Distorsiuni Interioare 11:17. The 80 to 20 split is one spoken estimate from a transcript, and no other figure turned up ecoustics 20:11. He has not said modern recordings are worse in every way: in 2018 he said he generally prefers modern quality recordings Head-Fi, Oct 2018, and in 2023 that modern good recordings "are better in every other respect" than early Decca Head-Fi, Aug 2023. His results with his own pulse array ADC are spoken descriptions from interviews and podcasts published in 2026; these items contain no published recordings or measurements from it Distorsiuni Interioare 11:57article, Oct 2026.

Quotes

“The recording quality with PCM I think is more important than the actual format.”2016-05-17 Head-Fi, May 2016

“the signal will have timing issues that the M scaler can't fix - once aliasing happens it is permanently damaged”2018-10-23 Head-Fi, Oct 2018

“Now the above deficiencies were entirely expected and I am convinced due to the ADC.”2022-03-16 Head-Fi, Mar 2022

“my estimate is around 80 percent of the sound quality losses is on the DAC side”2025-10-28 ecoustics 20:11

Dated positions

Caveats and open points

• Most of the technical statements about ADC aliasing and noise are generalizations about 'modern' or 'pro' ADCs; the converter he names for his 2022 recordings is the Focusrite Scarlett 18i8 (Head-Fi, Mar 2022).
• The 2022 recording post attributes the shortfalls to the ADC, but the same post says microphone position affected depth, and later transcripts say ribbon microphones can sound warm and that the recordings' warmth was partly the microphone (Head-Fi, Mar 2022, Soundnews 4:25, Distorsiuni Interioare 11:17).
• In 2022 he listed a warmer balance among the deficiencies he expects from the ADC (Head-Fi, Jul 2022); in a podcast published in October 2026 (recording date not stated) he said commercial ADCs are all too bright (Distorsiuni Interioare 1:31:20). This archive does not reconcile the two.
• The 80 percent and 20 to 25 percent figures are a verbal estimate; readers should check the recording (ecoustics 20:11).
• The cello distance is about 5 meters in the written March 2022 post but four meters in the transcripts of an interview published in August 2022 and a podcast published in October 2026 (Head-Fi, Mar 2022, AVstore 47:20, Part-Time Audiophile 14:32).
• His explanation of why older recordings sound as they do is hedged ('I suspect it may be bit of both', 'we can't be certain about this') (Head-Fi, Oct 2018, Head-Fi, Aug 2023).
• His 2026 reports on recordings made with his own pulse array ADC are spoken claims in items published in 2026; no published recordings or measurements from it appear in the items gathered (Distorsiuni Interioare 11:57, article, Oct 2026).
• Statements are scoped to the thread product where he named one; most are general statements about recordings, not about a particular Chord model.

Topics: Recordings, mastering and the production chain, Depth and soundstage, Decimation, aliasing and sample-rate conversion