Isn't pre-ringing bad? Why doesn't Chord offer minimum-phase, apodizing or NOS filters, and are NOS DACs really bit perfect?

Is pre-ringing bad?

For the DAC interpolation (reconstruction) filter, Watts's position has been the same from 2014 to 2026. In 2014 he wrote that "the high-end keeps banging on about the importance of no pre-ringing filters, completely the opposite to my thinking" Head-Fi, Apr 2014. His argument, repeated many times, is that an ideal sinc filter rings for ever in response to an impulse yet reconstructs a bandwidth-limited signal exactly, and that the paradox goes away because a test impulse is not bandwidth limited: it has as much energy at FS/2 as at DC, which sampling theory does not allow Head-Fi, Apr 2016Head-Fi, Jan 2017miceblue425 17:25Part-Time Audiophile 25:57. In 2016 he called ringing "a red herring" Head-Fi, Apr 2016. In December 2017 he wrote that the ideal interpolation filter, an infinite sinc, "categorically is not a slow roll off filter, but an ideal brick wall filter", and that the audio industry is "completely incorrect" in using non-bandwidth-limited impulses to infer the sound quality of digital systems Audiophile Style, Dec 2017Audiophile Style, Dec 2017.

He says that with a bandwidth-limited impulse a sinc filter shows no pre- or post-ringing, while shorter filters and minimum-phase filters do ring: "when you put a bandwidth limited impulse into a sinc function, it just does not ring at all" (Passion for Sound interview, Part 3, published December 2019; recording date not stated) Passion for Sound 15:30, and, in an interview published October 2026 (recording date not stated, though the transcript places it before May 2026), that "if you put in a bandwidth limited impulse, when you put it into a low tap -length filter, you get more ringing than when you put it into a long-tap-length filter" HiFi Buys 1:52:00. In a slide talk about the Blu Mk2 at Chord's Pump House, published by Hifistatement as audio clips in October 2017 (recording date not stated), he said of his own filters: "When you put an impulse response that is bandwidth limited into one of my filters, you get no extra ringing" Hifistatement 1:58. In April 2024 he restated that, with a bandwidth-limited impulse, longer sinc-function filters reduce pre- and post-ringing Head-Fi, Apr 2024. In 2025 he said a bandwidth-limited impulse "will show minimal pre and post ringing with a WTA filter, and as the filter gets longer, the less ringing it will show" Head-Fi, Mar 2025.

He goes further than calling it harmless: he says pre-ringing is how the filter uses future samples to place transients correctly. "So it's essential to have. Pre-ringing to reconstruct the transients properly" (RMAF 2017) miceblue425 19:34; "in order to correctly reconstruct the timing of transients the interpolation filter needs to know the future - and this creates the pre-ringing" (2022) Head-Fi, Jul 2022; "Without the pre-ringing, it's impossible to reconstruct the original" (Malaysia, 2025) Audiophile Junkie 7:18. He also argues any ringing would sit at FS/2, about 22 kHz for 44.1k, where nobody can hear it Head-Fi, Apr 2019Passion for Sound 19:20Head-Fi, Mar 2025. He has said the same of square waves: "If a DAC reproduces a square wave without ringing then it's time domain performance is an utter failure" (2023) Head-Fi, Dec 2023Head-Fi, Dec 2023.

He still treats the impulse test as useful for one thing: it shows a filter's coefficients, but it should not be used to judge sound quality miceblue425 17:25Head-Fi, Apr 2019HiFi Buys 1:51:31. In 2026 he added that impulse plots can mislead, since a long ringing filter followed by a notch filter would show no ringing HiFi Buys 1:55:08.

Where he says pre-ringing does matter, or may not

• EQ and other filtering inside the audio band. In 2019 he wrote that a filter with pre-ringing within the audio bandwidth "would definitely sound strange" Head-Fi, Apr 2019. In 2025: "if we are filtering in the audio domain, then pre-ringing is an issue", and "it's very important to use causal filtering or IIR for EQ", giving "the EQ in Mojo's UHD DSP" as an example of a causal IIR filter Head-Fi, Mar 2025.

• ADC decimation (the Davina ADC project). In 2014 he wrote that his listening tests had shown linear-phase brick-wall filters at the ADC input to be "completely transparent anyway, if you do them correctly" Head-Fi, Jul 2014. Later he said sampling theory does not care how the band limiting is done, so an IIR filter with no pre-ringing is acceptable in principle Head-Fi, Dec 2017Head-Fi, Sep 2018Head-Fi, May 2022. His expectation wavered before his 2022 report. In May 2016 he wrote "I actually suspect that pre-ringing is not an issue, that only aliasing is important" Head-Fi, May 2016 and "My current thinking is that ringing in an ADC will turn out to be unimportant" Head-Fi, May 2016; at RMAF in October 2017 he said "This might be a situation where pre-ringing is actually important" and planned listening tests miceblue425 41:33; in December 2017 he suspected that pre-ringing in the ADC bandwidth limiter "may not be a problem, but I will know soon with listening tests I have for the Davina project (my ADC)" Head-Fi, Dec 2017; and in 2018 and 2019 he guessed the FIR (pre-ringing) version would sound closer to transparent Head-Fi, Sep 2018Head-Fi, Apr 2019. In June 2021 he reported initial listening tests of his ADC bandwidth-limiting filters with "no evidence of any loss of transparency", calling his conclusions tentative Head-Fi, Jun 2021. In May 2022 he reported that his FIR decimation filters, listened to as plain low-pass filters, suggest "the pre-ringing is subjectively completely inconsequential" Head-Fi, May 2022. In 2017 he promised blind test files of FIR against IIR band limiting Head-Fi, Aug 2017; this archive does not show them published, and in a podcast published in October 2026 (recording date not stated) he still described decimation-then-interpolation comparisons with his ADC's 768 kHz recordings as tests to come Head-Fi, Aug 2017Part-Time Audiophile 1:03:53.

• M Scaler video mode. For low latency the Hugo M Scaler's video mode uses an asymmetric filter, short pre-ringing with the full taps for post-ringing; in 2023 he said "I originally listened to symmetric vs asymmetric with video mode and the asymmetric sounded a lot better" Head-Fi, Nov 2023Head-Fi, Jul 2018.

Why no minimum-phase, apodizing or NOS options?

His stated reason is that he judges them worse and will not ship options he knows degrade the sound. In Part 4 of the Passion for Sound interview, published in January 2020 (the series ran from December 2019; recording date not stated), he said offering minimum-phase and NOS options "would have taken no time whatsoever" and "If there's something that you know sounds worse I don't give that as an option to people" Passion for Sound 16:34. In May 2022: "I only give one version of the WTA1 filter", and he attributed other makers' filter menus to the view that "they do not understand what they are doing" Head-Fi, May 2022. In 2023, in the Dave thread: "For me to convert it to NOS or non-oversampling - zero effort", but it "would destroy musicality" Head-Fi, Nov 2023, and at CanJam SoCal 2023 he said he could put other filters in the interpolation filter but is "not doing that" The Source AV 37:04. In 2023 he also said that, given his responsibility to Chord, he would stay publicly silent if a rival filter were better, but "I would never, ever state that a filter sounded worse than mine, if that was not the case" Head-Fi, Aug 2023.

Minimum phase (including MQA's rendering filter). He has said minimum-phase filters lose the start of transients, a timing error he puts at about 100 microseconds in the transient illustrations he shows in talks (RMAF 2017, CanJam NYC 2022, Malaysia 2025) miceblue425 12:06MFG Audio Reviews 18:09MFG Audio Reviews 18:55Audiophile Junkie 5:21. In 2018: "I do not rate min phase filters at all", and he said they sound "flat, soft and ill defined" against WTA Head-Fi, Mar 2018. Also in 2018 he said MQA's filter sounded "very much worse than the WTA" and "there is no way I would ever offer this as an option" Head-Fi, Aug 2018; in May 2018 he said MQA's rendering creates transient timing errors "because of the mistaken and technically foolish belief that pre-ringing is bad" Head-Fi, May 2018; and he said that min-phase and NOS filters match an ideal sinc to only "1 bit accuracy" Head-Fi, Aug 2018. In the same Part 4 he called MQA "a minimum phase type filter" that is "only looking in the past" Passion for Sound 18:00.

Apodizing and slow roll-off. His wording differs between 2016 to 2017 and 2021 to 2023. In 2016 he wrote "Apodizing filters can offer better performance, but they still have substantial timing errors" Head-Fi, Oct 2016, and in 2017 that a slow roll-off or apodizing filter about -30 dB at FS/2 "will sound better than a half band pseudo brick wall filter", though a brick-wall filter with sensible attenuation "will easily beat" it Head-Fi, Oct 2017. From 2021 to 2023, discussing PGGB, he said apodizing will not reconstruct transients properly whatever the tap length Head-Fi, May 2021Head-Fi, Jun 2021Head-Fi, Jun 2021Eric's Hi-Fi 11:13; in 2022: "It doesn't matter if it's 10 taps or 10 billion taps - apodizing is the wrong approach" Head-Fi, Jul 2022. He names two aims of apodizing, removing pre-ringing and suppressing ADC aliasing between 20 and 22.05 kHz, and reported that filtering that band after WTA1 at 705 kHz gave "little or no sound quality benefits from attenuating ADC aliasing" Head-Fi, Jul 2022.

NOS. In December 2017: "NOS DAC's - if your intent is to be as transparent as possible - are an unmitigated disaster", sounding "soft, flat and out of focus" Head-Fi, Dec 2017. He describes NOS as the crudest interpolation filter, with transient timing errors of up to about 100 µs Head-Fi, Oct 2016Head-Fi, Jul 2019HIFICLUB 29:50 and sampling images extending to infinite frequencies Head-Fi, Dec 2017. He has repeated this through 2026 Part-Time Audiophile 39:00HiFi Buys 42:01.

Why people like these filters, in his account. He says the timing errors soften the sound, which can mask the hardness of a conventional DAC (noise floor modulation in his 2022 CanJam NYC talk, high-frequency distortion in the interview published October 2026, recording date not stated): "on inferior DACs, people could prefer the sound of a NOS filter" (Passion for Sound Part 4, published January 2020; recording date not stated) Passion for Sound 21:48MFG Audio Reviews 18:55HiFi Buys 42:01. In May 2022 he wrote that min phase, MQA, apodizing and NOS filters "seriously degrade transient timing", and that the desire for them "is driven by conventional DACs sounding hard and aggressive" Head-Fi, May 2022. In 2019 he wrote that he did not know whether the industry's "ringing is bad" view comes from technical ignorance or marketing, and suspected both Head-Fi, Apr 2019.

The filter options Chord does offer

Hugo 2, Hugo TT2 and Qutest have selectable filters, which he described in 2017 as "a big departure for me" Head-Fi, Mar 2017. The 16FS (orange or red) setting switches out the 16FS to 256FS WTA2 stage; in 2017 he said "the 16 FS option is just IIR filters (analogue type filter)" Head-Fi, Jul 2017, and in 2018 that it "gets replaced with an IIR filter (similar to minimum phase in no pre-ringing) and again it sounds warmer and softer with the timing imprecision" Head-Fi, Jul 2018. In a January 2025 post in the Qutest thread that does not name the product, he said WTA1 is unaffected by the options Head-Fi, Jan 2025. He said he included the option to let people hear what WTA2 does and to tame overly bright headphones or recordings, while calling the 256FS setting "the technically more accurate filter" and the option one that "very much goes against my purist design philosophy" Head-Fi, Mar 2017Head-Fi, Jul 2017. Dave runs the 256FS WTA stage all the time Head-Fi, Jul 2018. Among Chord's filter settings, it is the only one he links to no pre-ringing, and it applies above 16FS, after WTA1 Head-Fi, Jul 2018Head-Fi, Jan 2025.

Are NOS DACs bit perfect?

No, in his account. In 2019 he wrote: "bit perfect NOS DACs are NOT bit perfect as sampling theory holds that the samples are of zero width, and in between samples is a virtual zero", and that NOS replaces those samples with a sample-and-hold filter with "huge timing errors with transients" Head-Fi, Mar 2019. He said "I much prefer the concept of signal perfect", meaning faithful to the original analog signal in the ADC, which he says the M Scaler achieves to at least 16-bit precision Head-Fi, Mar 2019. In 2015 he called the idea of a bit-perfect DAC ridiculous, since a DAC's job is to reproduce the continuous waveform rather than the sampled data with its errors intact Head-Fi, Apr 2015. He makes the same point when saying a DAC's job is "NOT to reproduce the digital data" (2015) Head-Fi, Oct 2015 and when discussing NOS at a 2018 seminar MajorHiFi 2:37.

He also says the label "non-oversampling" is wrong, because every DAC oversamples and NOS is itself a digital oversampling filter: "the term NOS for a DAC is a non sequitur or a delusion" (2023) Head-Fi, Nov 2023Head-Fi, Jul 2019Soundnews 14:48ecoustics 22:30. He has named it inconsistently: "zero order hold" in 2016 and 2019 posts Head-Fi, Oct 2016Head-Fi, Jul 2019, and "first order hold" in 2023 to 2026 posts and talks Head-Fi, Nov 2023HIFICLUB 28:56Part-Time Audiophile 39:00. In 2015 he noted that common half-band filters also "retain untouched the original samples", and that his WTA filter does not, by design Head-Fi, Jun 2015.

This is separate from source data. He routinely asks that the data reaching a Chord DAC be bit perfect, with no upsampling or gain changes in the player Head-Fi, Apr 2018Head-Fi, Mar 2016Head-Fi, Dec 2016.

What he has not said: there is no published measurement from him comparing a bandwidth-limited impulse through WTA against a minimum-phase filter (in 2016 he said he hoped to publish time-domain measurements Head-Fi, Mar 2016), and no published results of the FIR against IIR decimation files he promised Head-Fi, Aug 2017. He has said he will not offer NOS or minimum-phase options Passion for Sound 16:34Head-Fi, Nov 2023.

Quotes

“If you use bandwidth limited impulses then you will get no pre-ringing, and a sinc function filter will always perfectly reconstruct the original analogue signal.”2022-07-15 Head-Fi, Jul 2022

“If there's something that you know sounds worse I don't give that as an option to people.”2020-01-02 Passion for Sound 16:34

“…bit perfect NOS DACs are NOT bit perfect as sampling theory holds that the samples are of zero width, and in between samples is a virtual zero…”2019-03-29 Head-Fi, Mar 2019

“…if we are filtering in the audio domain, then pre-ringing is an issue, because pre-ringing artifacts would be in the audio bandwidth and hence audible…”2025-03-26 Head-Fi, Mar 2025

Dated positions

Caveats and open points

• The roughly 100 microsecond timing-error figure for NOS and minimum-phase filters comes partly from a 2016 post (Head-Fi, Oct 2016) and mostly from talks that use a tone-burst illustration; in 2017 he said a similar Mojo illustration was deliberately simple and not bandwidth limited (Head-Fi, Feb 2017). Check the recordings.
• He names the NOS filter both 'zero order hold' (2016, 2019 posts) and 'first order hold' (2023 to 2026); the answer reports both without resolving the terminology.
• His statements on ADC decimation pre-ringing are tentative listening results from the unreleased Davina project, not a published test.
• The Hugo 2 / TT2 / Qutest 16FS option and M Scaler video mode are the only Chord filters in this archive he links to reduced or absent pre-ringing; the answer does not assess them beyond his words.
• Several of his strongest statements about rival designs are opinions about competitors' products and motives, reported as his views.

Topics: Pre-ringing, phase and rival filter designs, Sampling theory and ideal reconstruction