• Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Okto DAC8 update with DSP

OK, I can well understand your above preference. ;)

On the other hand...
Just for your possible reference and interest, at least in my PC-DSP-based multichannel multi-SP-driver multi-amplifier full active audio system (details can be found in #931, #1,009, #1,022 on my project thread), and as far as "my present ears and brain" concern, "on-the-fly (real-time)" conversion (over-sampling, down-sampling) of all the digital tracks into 88.2 kHz or 96 kHz PCM by JRiver MC is more than enough, and it is even needed.

You can find the "historic sequences", background, reasons and rationales in my post #532 on my project thread.
_
Enlightening read, thank you @dualazmak !

One does wonder, however, about the trajectory of the "sufficient bit depth and frequency for faithful reproduction of sound"... If memory serves, 44100hz used to be the "final frontier" of sorts -- much like the famous "you'll never need more than 128kb of RAM".

DSP is another thing I'm not quite ready to get into; my listening space is way too unoptimized as it is, and I don't believe DSP is a solution for that

My current _working_ solution is just to keep using the windows PC (ms surface 4, so not really good for anything else these days) and okto's ASIO driver. It just works (native DSD, whatever channels). I just got tired of chatting with AI about it, so decided to go "human" -- so thank you for indulging!
 
Hi Alex,
yes you found the reason, this is because the DAC8PRO propose "only" 3 alternate settings for DSD 2/6/8
This is already a major difference with all other device on earth which propose only one DSD format for their number of channels (say 8 for example)
in the last software version I have managed to bundle a total of 8 alternate settings to try covering almost all format (5 pcm, 3 dsd) and I did consider 8 being a kind of reasonable effort (considering the work of copy paste needed to do this in the original xmos sw stack) and this was typically for being compliant with Alsa.
Alsa always try to find an alternate aligned with the source. Unless using a "plughw" . But plughw does not support dsd format.
So as suggested above the Player should find way to accomodate by converting 5.0 to 5.1.
adding a 9th alternate would be feasible in a next FW version, but then why not supporting 4.0 (ambiphonic) or 2.1 and so on, so I will not promise anything here today , and I hope you will find a solution to overcome this :)
 
Hi Alex,
yes you found the reason, this is because the DAC8PRO propose "only" 3 alternate settings for DSD 2/6/8
This is already a major difference with all other device on earth which propose only one DSD format for their number of channels (say 8 for example)
in the last software version I have managed to bundle a total of 8 alternate settings to try covering almost all format (5 pcm, 3 dsd) and I did consider 8 being a kind of reasonable effort (considering the work of copy paste needed to do this in the original xmos sw stack) and this was typically for being compliant with Alsa.
Alsa always try to find an alternate aligned with the source. Unless using a "plughw" . But plughw does not support dsd format.
So as suggested above the Player should find way to accomodate by converting 5.0 to 5.1.
adding a 9th alternate would be feasible in a next FW version, but then why not supporting 4.0 (ambiphonic) or 2.1 and so on, so I will not promise anything here today , and I hope you will find a solution to overcome this :)
thank you @fabriceo !
I do realize how much effort goes into developing something like this, and I am grateful for yours :)
plughw: that's how JRMC _can_ play dsd 5.0 (via DoP)
in defense of my plea to prioritize DSD5.0 is that there are A LOT of releases of classical music in this format specifically (well, 5.0 SACD...)
anyways, will keep my fingers crossed (and stay with Windows/ASIO for now).
Cheers,
A
 
Ok, a bit of an update.

Disclaimer: I am not a programmer, just an enthusiast willing to play ;)

After poking around JRMC and ALSA, I asked my AI helpers if looking at a different player -- MPD, in this instance -- could yield different results. I knew from before (from my playing with a PC install of Volumio) that MPD too could play stereo dsf without issues, 5.1 dsf with a hiss, and choke on 5.0 dsf.

Claude wrote a Python script inserting a silent LFE channel in slot 3 (so, 0x69; first slot being 0), and I asked Claude if it were possible to run this code on the fly before the player hands over to ALSA. Claude said "sure", wrote an initial script to patch the MPD source files, and bowed out (I use free tier on both -- but after this adventure I am sure I will get a subscription).

ChatGPT entered the stage, and "we" worked on this most of the whole day yesterday, with diagnosing, capturing the data at various stages, comparing capture to originals, reversing bytes, rebuilding MPD, etc, etc, etc.

The insertion of the silent LFE was the easiest part (or, I should say, took the least amount of time to implement and troubleshoot). The main issue at hand was dsd-to-pcm conversion that MPD was doing _somewhere in the chain_ without being asked, and ChatGPT was kind of softly refusing to look at it with "but the code here says it should not be happening, and we have no evidence of this, and lets look at this other part first and find out who is calling the function" and so on round and round, through multiple context windows. Eventually, "we" looked at that, and discovered that the conditional dsd2pcm conversion was happening way upstream, even when the conditions were not calling for it -- proven by multiple stream captures and analyses.

Few more diagnostic runs, and patches were inserted into relevant files, MPD rebuilt -- and voila! It plays correctly now, tested with 2.0 DSD64 and DSD128, 5.0 DSD64, original 5.1 DSD64; stereo and MCh flac (96 to 192kHz / 24 to 32 bit, to make sure no shenanigans with channel fiddling was happening). The DAC shows proper format playing, and proper channels active.

I will be asking ChatGPT to compile a bug report for the MPD ppl ('cuz I have no clue how this is done).

MiniPC, Ubuntu 26.10, MPD 0.24.12

Cheers all,

Alex
 
good that you solved it, patching MPD with help of AI is very inspiring... and amazing...
In a same way a good exercise would be to patch alsa and especially the plughw plugin, so that it accepts the DSD32 format (in a passthrough approach) and just realign the number of incoming channels to the target max channels, what it does in pcm.

anyway, thanks for sharing
 
Back
Top Bottom