You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
DMIC capture accepts 4ch recording but gives corrupted audio when configured for 2ch #327
The number of channels (used PDM stereo controllers count) is defined in topology parameters. Now the capture runs but results corrupted audio. Example of this kind of violation is to use topology sof-apl-nocodec.tplg or sof-apl-da7219.m4 and run capture like:
The time print shows that such capture will take 20s instead of requested 10. The 2ch audio samples will be distributed to all 4ch and recording will sound double speed and double pitch.
The proper driver behavior should be to reject capture of non-matching channels count.
changed the title [-]DMIC capture accepts 4ch recordin but gives corrupted audio when configured for 2ch[/-][+]DMIC capture accepts 4ch recording but gives corrupted audio when configured for 2ch[/+]on Sep 7, 2018
@singalsu ok so dmic driver should reject channels if not supported by topology. Host driver does not have internal pipeline logic to do this rejection.
@michalgrodzicki@lgirdwood@juimonen No, SOFT PR #90 is not. I feel it could be useful for merge if #92 that has a wider scope takes still time to develop. Though #90 seems to need an update. What is your recommendation? This has been open for a long time with first simple fix #90 on hold.
Activity
[-]DMIC capture accepts 4ch recordin but gives corrupted audio when configured for 2ch[/-][+]DMIC capture accepts 4ch recording but gives corrupted audio when configured for 2ch[/+]lgirdwood commentedon Sep 7, 2018
@singalsu ok so dmic driver should reject channels if not supported by topology. Host driver does not have internal pipeline logic to do this rejection.
singalsu commentedon Oct 12, 2018
As update the fix is under construction in SOFT PR #90.
michalgrodzicki commentedon Oct 17, 2018
@singalsu PR #90 is closed, what is a status of this issue?
singalsu commentedon Oct 18, 2018
@michalgrodzicki @lgirdwood @juimonen No, SOFT PR #90 is not. I feel it could be useful for merge if #92 that has a wider scope takes still time to develop. Though #90 seems to need an update. What is your recommendation? This has been open for a long time with first simple fix #90 on hold.
lgirdwood commentedon Oct 18, 2018
@singalsu merged now, so @juimonen can fix properly after his current commitment.
singalsu commentedon Oct 22, 2018
This issue can be closed. I just checked that arecord command with four channels is rejected (and two channels works).
singalsu commentedon Oct 22, 2018
Closing.