- Joined
- Apr 24, 2026
- Messages
- 16
- Likes
- 12
Greetings! I am Magos Vox; some of you may have crossed paths with Vox Machina, a bit-perfect USB-DAC engine for Android. I come bearing data.
I have recently implemented a feature I cautiously suspect this room in particular will appreciate: a voluntary transmission, back to the Forge, of the USB descriptors emitted by a listener's attached DAC - the device's own sworn statement of what it will accept. I collate these, and set the manufacturer's declared specifications against the truth of the silicon.
I call it the Concordance.
A clarity owed at the outset: this concerns what a device declares it can do, and whether it then does it - not how it measures up once the music plays. The analog truth is your altar, and you hold it better than anyone. Mine is the narrower question: does the silicon's own word match the words on the box?
A few cases have already surfaced.
The Concordance was born, fittingly, of a complaint. The Engine refused to render DSD natively through a listener's JadeAudio JA11. He had trusted listings and images selling it as natively DSD128-capable - yet the silicon told another story: no native DSD path at all; only PCM, with DSD carried over DoP, and no further. The marketing promised a door the descriptor does not open. One caveat: the descriptor witnesses what a device declares - and a handful of DACs do native DSD without declaring it the standard way. UAC2 defines no DSD format of its own; the flag everyone leans on is a de-facto convention, and older/proprietary firmwares simply skip the flag - the DSD alt then looks like ordinary PCM on the wire or gate native DSD behind a proprietary driver entirely. The JA11 isn't one of them; its maker declares native DSD when present, so here the silence is conclusive.
The happier kind is just as instructive. A few DACs sold as native DSD256 proved - in silicon, and in play - to render DSD512 natively, without complaint. The box undersold what the device will gladly do.
That second sort matters as much as the first. The Concordance is not a gallows; it reports the gap whichever way it runs, and now and again the silicon turns out more honest than the brochure that sells it.
In time, the hope is this builds up into something genuinely useful - a reference for what these devices actually declare, gathered from the field rather than the marketing department.
P.S. The Concordance is entirely voluntary. The payload carries only the attached DAC's descriptor - model name, USB vendor/product IDs, and audio profiles - plus a random token to discard duplicate submissions. Nothing about the phone, OS, or the listener is read or included.
I have recently implemented a feature I cautiously suspect this room in particular will appreciate: a voluntary transmission, back to the Forge, of the USB descriptors emitted by a listener's attached DAC - the device's own sworn statement of what it will accept. I collate these, and set the manufacturer's declared specifications against the truth of the silicon.
I call it the Concordance.
A clarity owed at the outset: this concerns what a device declares it can do, and whether it then does it - not how it measures up once the music plays. The analog truth is your altar, and you hold it better than anyone. Mine is the narrower question: does the silicon's own word match the words on the box?
A few cases have already surfaced.
The Concordance was born, fittingly, of a complaint. The Engine refused to render DSD natively through a listener's JadeAudio JA11. He had trusted listings and images selling it as natively DSD128-capable - yet the silicon told another story: no native DSD path at all; only PCM, with DSD carried over DoP, and no further. The marketing promised a door the descriptor does not open. One caveat: the descriptor witnesses what a device declares - and a handful of DACs do native DSD without declaring it the standard way. UAC2 defines no DSD format of its own; the flag everyone leans on is a de-facto convention, and older/proprietary firmwares simply skip the flag - the DSD alt then looks like ordinary PCM on the wire or gate native DSD behind a proprietary driver entirely. The JA11 isn't one of them; its maker declares native DSD when present, so here the silence is conclusive.
The happier kind is just as instructive. A few DACs sold as native DSD256 proved - in silicon, and in play - to render DSD512 natively, without complaint. The box undersold what the device will gladly do.
That second sort matters as much as the first. The Concordance is not a gallows; it reports the gap whichever way it runs, and now and again the silicon turns out more honest than the brochure that sells it.
In time, the hope is this builds up into something genuinely useful - a reference for what these devices actually declare, gathered from the field rather than the marketing department.
P.S. The Concordance is entirely voluntary. The payload carries only the attached DAC's descriptor - model name, USB vendor/product IDs, and audio profiles - plus a random token to discard duplicate submissions. Nothing about the phone, OS, or the listener is read or included.
