Do lossy or streamed files sound worse than local files on a Chord DAC?

Lossy files (MP3, AAC)

• What reaches the DAC. In 2017, on Hugo 2, he wrote that the filters "can't possibly know whether the PCM was originally MP3, AAC or straight uncompressed PCM" Head-Fi, Jul 2017, and that "AAC is fine, it's decoded by the source", unlike Dolby and DTS data Head-Fi, Sep 2017.
• 2014 to 2016: Hugo, Mojo, Dave. He was positive. In 2014 he was "constantly amazed at how good AAC sounds" on Hugo Head-Fi, Apr 2014. In 2015 he was "constantly surprised how good compressed AAC can sound on Hugo" Head-Fi, Apr 2015, and in 2016, of Mojo, Hugo and Dave: "my surprise is how good compressed music can sound" Head-Fi, Apr 2016.
• 2017 to 2019: longer filters, M scaler, TT2. At CanJam RMAF in October 2017 he said "AAC sounds soft by comparison" with regular 16-bit and that he can "easily hear it using a WTA tap length filter" miceblue425 58:36. In April 2018 he wrote that MP3 or AAC 320 sounds "very different than redbook" if the recording was good, and "you can't mistake lossy compression on a BluDave for example, but is less easy to hear on a regular DAC" Head-Fi, Apr 2018. He hedged: on tracks meant to prove compression makes no difference, "I could not reliably tell the difference; but all the tracks sounded poor, even the so called bit perfect one" Head-Fi, Apr 2018. With the Hugo M Scaler, "it becomes much easier to hear the losses due to compression" (August 2018) Head-Fi, Aug 2018; the softness of 320k MP3/AAC is "much more apparent via TT2 and M scaler" (November 2018) Head-Fi, Nov 2018; and an M scaler makes MP3 deficiencies "much more obvious" (March 2019) Head-Fi, Mar 2019.
• His own shift. In July 2018 he said he had liked AAC's warm presentation "For many years (pre Hugo)", but "Although I used to enjoy AAC, I no longer can", because redbook via an M scaler is warm and smooth enough Head-Fi, Jul 2018. In May 2021 he wrote "I think that's why MP3 sounds soft compared to redbook", referring to transients not being encoded with great accuracy Head-Fi, May 2021.

Streaming services versus local files

• Same recording, same sound. In an interview published in February 2025 and recorded at a Munich show (most likely May 2024, unverified) he said "I use Qobuz and if the recording is the same recording as the one I've got on my hard disk, in practical terms they sound the same" audio-2G 12:31. He keeps his listening tests local because with streaming "you don't know the source, you don't know whether they've changed it" audio-2G 11:57. Asked in the same interview whether network gear such as routers, switches and cables affects the sound (the question is cut off), he answered "So yeah, it does" audio-2G 11:41audio-2G 11:52.
• Which version you get. "You can never be sure with streaming what they're giving you" audio-2G 12:43. In August 2023 a 48k 24b Qobuz copy of an album he owned at 16b 44.1k "sounded like a completely different album", which he said shows "how much damage non integer sample rate conversion from 48k to 44.1k does" Head-Fi, Aug 2023. In 2021, about a streaming app that seemed to be upsampling, he said "You need to ensure that the app gives you the original sample rate of the file - that's the reason I show the incoming sample rate" Head-Fi, Aug 2021.
• He streams himself. In September 2020 he wrote "I haven't tried streaming - as I like to own the music" Head-Fi, Sep 2020. By 2022 he was listening on Qobuz Head-Fi, Sep 2022; in 2023 he said he runs Qobuz from a PC over optical Head-Fi, Nov 2023; in July 2024: "Today, my new music comes via Qobuz as PCM" Head-Fi, Jul 2024.
• The streamer can still matter. In February 2023 he said bit-perfect sources should sound identical, but "Currently, with my DACs, they do not sound the same. You will have different streamers sounding different" Dusty Audio 48:52. In an interview published in May 2026 (recording date not stated) he described making a poor and a good streamer sound the same as his intent: "And this I think we can do" Audio Force 18:19.
• Send the file unaltered. In November 2024: "In other words always send bit perfect data", with no EQ, volume or upsampling by the source Head-Fi, Nov 2024.
• Tidal and MQA. In 2018: "My advice is to ignore MQA and always go for the unchanged original file" Head-Fi, Aug 2018. Asked in February 2023 about MQA on Tidal, he said "my advice is don't use Tidal" Dusty Audio 57:29.

Wireless links

In 2017 he noted that aptX HD "is still a lossy format even at 48 kHz" Head-Fi, Sep 2017. In 2022 he said "I would never use a lossy WiFi connection to transmit audio" even for subwoofers, meaning a link that adds an extra ADC and DAC Head-Fi, Apr 2022. There is no statement from him on lossy files for Qutest, Mojo 2 or Quartet, on Spotify or Apple Music quality tiers, or on a bitrate at which lossy becomes inaudible.

Quotes

“But a MP3 or AAC 320 sounds very different than redbook, if the original recording was good, with the compressed sounding softer and warmer.”2018-04-14 Head-Fi, Apr 2018

“Although I used to enjoy AAC, I no longer can”2018-07-01 Head-Fi, Jul 2018

“I use Qobuz and if the recording is the same recording as the one I've got on my hard disk, in practical terms they sound the same.”2024-05 audio-2G 12:31

“In other words always send bit perfect data.”2024-11-11 Head-Fi, Nov 2024

Dated positions

Caveats and open points

• His comments on lossy sound are listening impressions; in the 2018-04 post he also says he could not reliably tell the difference on some test tracks.
• His 2014 to 2016 positive comments on AAC and his 2017 to 2021 comments on softness differ in product and setup; he describes his own shift in the 2018-07-01 post.
• The 2025 answer on network gear responds to a question that is cut off in the transcript.
• No statement was found on Spotify or Apple Music quality tiers, or on Qutest, Mojo 2 or Quartet with lossy files.
• The CanJam 2017 pre-echo comment about 320 kbps was made by an audience member, not Watts.
• The audio-2G interview was published on 17 February 2025; it was most likely recorded at a Munich show in May 2024 (date unconfirmed).
• The in_his_words date 2024-05 for the Qobuz quote is the likely month of the Munich show (unverified); the interview was published 2025-02-17.

Topics: MQA, lossy codecs and streaming services, Source devices, streamers and networks