Do Chord DACs decode MQA, and what does Rob think of it?

Do Chord DACs decode MQA?

2016: under evaluation. In August 2016, in the Dave thread, Watts wrote "MQA is still under evaluation", adding that Dave's FPGA and spare expansion connector meant Chord could support future developments "if I was convinced that it offered a credible improvement in sound quality" Head-Fi, Aug 2016.

2017 onward: refused. By May 2017 he had decided. In the Dave thread he said he had compared the MQA interpolation filters against a standard WTA filter using a Mojo, that MQA "is technically flawed, with huge and completely unacceptable levels of aliasing", and "it's those two reasons why I refuse to implement it in my designs" Head-Fi, May 2017. In August 2018, on the MQA interpolation filter, he wrote "there is no way I would ever offer this as an option" Head-Fi, Aug 2018. In a February 2023 talk in Bucharest he recalled listening to the MQA filters when they first came out, hearing "enormous degradations", and saying "no, we're not doing it", at a cost: "a lot of customers would not buy our DACs because we took that stand" Dusty Audio 55:44.

Decoding in software instead. In the Hugo M Scaler thread in August 2018 he said MQA's first unfold "is best done in your media player, which Tidal and Roon can both do", while adding "You are still better off with the original undamaged 44.1" Head-Fi, Aug 2018. Two days later, explaining the M Scaler's sample rate colors, he wrote "If it's yellow, then it's 88.2, which for MQA would mean the unfold has been done", and again preferred the original 44.1k "IMHO" Head-Fi, Aug 2018. At an April 2022 Mojo 2 interview in Thailand he said "that's the reason we do not support MQA" and "if you have got an MQA recording you can decode it in your source", so "not having the decoding available in in Mojo 2 is not a disadvantage" Audio Force 12:40Audio Force 13:28. Asked in Bucharest in February 2023 how Dave and the M Scaler handle MQA files, he said "You can, of course, decode the MQA through your software", mentioned reports that some people prefer the files undecoded while saying "I haven't tried it myself", and, when the questioner raised Tidal, said "my my advice is don't use Tidal" Dusty Audio 56:46Dusty Audio 57:29.

What he thinks of MQA

2017: listening comparisons

In May 2017, in the Hugo 2 thread, he said the MQA interpolation filters, "with their huge timing errors due to aliasing, sounded dreadful in comparison to the WTA interpolation filter", and hedged: "Of course I am biased, but just don't make the assumption that MQA is better and listen for yourself" Head-Fi, May 2017. In November 2017 he said the problems are audible on a Mojo but more apparent on Hugo 2 because of its extra transparency and better WTA filters Head-Fi, Nov 2017. In February 2017 he referred to "the poor aliasing performance of MQA and other conventional filters" Head-Fi, Feb 2017.

He has given two different accuracy figures. In January 2017 he wrote that "a conventional interpolation filter (and MQA) is about 6 to 8 bits accurate for the interpolated samples with transients" Head-Fi, Jan 2017. In an audio clip about the Blu MkII, a slide talk at Chord's Pump House published by Hifistatement in October 2017 (recording date not stated), he is transcribed as saying "MQA is three or 4 bits accurate" Hifistatement 24.

2018: a lossy codec with measured faults

In May 2018 he wrote that "technically the format is clearly a lossy distortion generating codec", pointing to published third-party measurements (1% THD and noise at 20 kHz, a notch at 44.1k) and to his worry that MQA rendering "actually creates transient timing errors, because of the mistaken and technically foolish belief that pre-ringing is bad" Head-Fi, May 2018. In June 2018 he said MQA's "leaky filters have appalling aliasing performance of around -40dB", that 1% distortion at 20 kHz "is simply broken, and not fit for purpose", and that the 1% figure came from an independent blogger's measurements rather than his own Head-Fi, Jun 2018. The same day he wrote "The MQA sound does depend upon the DAC you use", and that its degradations become very obvious with a WTA filter, particularly with an M Scaler Head-Fi, Jun 2018.

His most detailed written breakdown is from August 2018. He split MQA into the unfold and a "low tap length minimum phase interpolation filter". Against his WTA filter, he said, the MQA filter sounds "soft, fat bloated bass, poor instrument separation and focus, with a flat ill defined sound-stage". Of the compression stage he cited a notch at 22.05 or 24 kHz, aliasing giving 1% distortion at 20 kHz, and lossiness that "converts a 24 bit signal into something like 17 bits". He ended: "My advice is to ignore MQA and always go for the unchanged original file" Head-Fi, Aug 2018. In July 2018 he grouped "minimum phase, NOS, MQA on the DAC side" as technical disasters for image aliasing Head-Fi, Jul 2018, and tied the soft, bloated bass of MQA, minimum phase and NOS filters to transient timing uncertainty Head-Fi, Jul 2018.

2019 to 2020: filter theory interviews

In April 2019 he dismissed peer review with the remark that "MQA was so called peer reviewed" Head-Fi, Apr 2019. In a Passion for Sound interview published in December 2019 (recording date not stated) he said "The industry, and that includes MQA, have got this completely wrong" Passion for Sound 12:56, and that a bandwidth-limited impulse put into "an MQA filter or a minimum phase filter" will ring more than it should, unlike a sinc function Passion for Sound 15:30. In part 4, published January 2020 (recording date not stated), he said "MQA, again, is a minimum phase type filter", poor because "it's low tap length" and because it looks only into the past, and that calling pre-ringing bad "is a prejudice" Passion for Sound 18:00. He explained listener preference for MQA or minimum phase as a way of taming a DAC that sounds hard and bright from noise floor modulation: transients go out of focus, so the sound becomes warmer and the bass "big and fat and ill defined" Passion for Sound 20:31, so that on inferior DACs "it's artificially softening the sound" Passion for Sound 21:48.

2020 to 2025: not fit for purpose

In September 2020 his rule of thumb for buying music excluded processed versions: "certainly not the distorted MQA version, nor an upsampled or re-processed version" Head-Fi, Sep 2020. In April 2021 he wrote "I knew MQA was a potential disaster from the AES papers, and this was confirmed after listening tests on the MQA filters", and repeated that aliasing of -40dB is "completely unacceptable" Head-Fi, Apr 2021. At CanJam NYC 2022 (late February; the video was published in March 2022), illustrating how a tone burst's start is reconstructed, he grouped "an MQA filter" with "all these minimum phase nonsense filters" MFG Audio Reviews 18:09; he had used the same illustration, naming MQA filters alongside NOS and IIR filters, at CanJam London in July 2018 and at the Nintronics Open Day in November 2018 (uploaded December 2018) c w 9:57Pursuit Perfect System 5:23. In the April 2022 Thai interview he said "The problem with MQA is that it destroys transient information because the process has 1% worth of aliasing distortion" Audio Force 12:40, and that "MQA is fundamentally the the wrong way to go forward" for his goal of the sound of live unamplified instruments Audio Force 13:28. In May 2022 he grouped "min phase/MQA/apodizing/NOS filters" as seriously degrading transient timing Head-Fi, May 2022. In a November 2022 reply he wrote "Very badly... Of course it depends upon how sensitive you are as a listener... My view is that MQA is not fit for purpose" Head-Fi, Nov 2022.

In Bucharest in February 2023 he repeated the case: very limited taps, aliasing, and measurements by others showing "1% distortion above 20 kHz", which he called "completely unacceptable" Dusty Audio 55:01. He again explained the appeal as lost transient information giving a bass that is "fat and flabby and lacking tunefulness" that some people like Dusty Audio 58:14. In a May 2025 talk in Malaysia he listed "minimum phase filters, and MQA filters" among the simple interpolation filters with a transient error of around 100 microseconds Audiophile Junkie 5:21. The figures should be checked against the recordings.

Hedges he attached

He has qualified his view several times: "Of course I am biased" (May 2017) Head-Fi, May 2017; "IMHO" on preferring the original 44.1k (August 2018) Head-Fi, Aug 2018; "it depends upon how sensitive you are as a listener" (November 2022) Head-Fi, Nov 2022; and, on playing MQA files undecoded, "I haven't tried it myself" (February 2023) Dusty Audio 56:46. The 1% distortion figure is, by his own account, from third-party measurements, not his own Head-Fi, Jun 2018Dusty Audio 55:01.

What he has not said. His refusals are stated generally, as "my designs" (2017) and "we do not support MQA" (2022) Head-Fi, May 2017Audio Force 12:40. The product-specific statements cover Dave (2016, under evaluation), the Hugo M Scaler's handling of software-unfolded MQA (2018), Mojo 2 (2022) and Dave and the M Scaler (2023) Head-Fi, Aug 2016Head-Fi, Aug 2018Audio Force 13:28Dusty Audio 56:46. He does not address Qutest, Hugo TT2 or Poly by name on this question. The listening he describes is of MQA's interpolation filters, on a Mojo or fed directly to the DAC Head-Fi, May 2017Dusty Audio 55:44; he does not say whether he has heard a full MQA decode-and-render chain on another maker's hardware.

Quotes

“MQA is technically flawed, with huge and completely unacceptable levels of aliasing. And it's those two reasons why I refuse to implement it in my designs.”2017-05-18 Head-Fi, May 2017

“My advice is to ignore MQA and always go for the unchanged original file”2018-08-10 Head-Fi, Aug 2018

“Of course if you have got an MQA recording you can decode it in your source.”2022-04-26 Audio Force 13:28

“My view is that MQA is not fit for purpose. 1% HF distortion is simply not Hi-Fi let alone high end audio.”2022-11-23 Head-Fi, Nov 2022

Dated positions

Caveats and open points

• His refusals are general ("my designs", 2017; "we do not support MQA", 2022), and product-specific statements cover Dave (2016), the Hugo M Scaler (2018), Mojo 2 (2022) and Dave and the M Scaler (2023).
• The 1% distortion at 20 kHz figure is, by his own account, from third-party measurements, not his own (Head-Fi, Jun 2018, Dusty Audio 55:01).
• He gave different bit-accuracy figures for MQA filtering: 6 to 8 bits in a January 2017 post (Head-Fi, Jan 2017) and three or four bits in an audio clip published October 2017 (recording date not stated) (Hifistatement 24).
• His Bucharest remark that some people prefer MQA files undecoded was based on reports he had not tried himself (Dusty Audio 56:46).

Topics: MQA, lossy codecs and streaming services, Pre-ringing, phase and rival filter designs