Seppo Ingalsuo
Seppo Ingalsuo
I noticed this issue accidentally when working with a development version testbench that failed in topology parse.
> @singalsu @andrula-song this PR uses pass-through coef. Please help to figure out a non-pass-through coef case. Thanks! Thanks! I will add the Matlab scripts to generate the test vectors...
This is tricky, need a cmake expert here.
> @singalsu is this something you can self assign and fix (maybe via testbench or plugin if ready/faster ?) I'll check with @btian1 his findings, if no solution before CNY...
The issue can be repeated in playback side with a simple control sequence: ``` amixer cset name='Post Mixer Analog Playback Volume' 45 sleep 1 amixer cset name='Post Mixer Analog Playback...
Fix work in progress: https://github.com/thesofproject/sof/pull/8950
Fixed by #8950.
> good finding! we lack volume change test There is volume_check_levels.sh in sof-test for nocodec topologies. I've just set improvements PR to it to be more robust against control timing...
@plbossart We want to keep capture DAI 32 bits even if capture format to user space is 16 bits. We can't provide to user space 16 bit quality with 16...
> Okay, this makes sense @singalsu, but the commit message doesn't even begin to explain how this is related to 16 bits and that the choice of 32 bits is...