Tools: Topology2: Add pre-processing to ACE 1/2/3.x PCH DMIC topologies #9690
Conversation
The topologies for PCs those have PCH DMIC enabled have too silent noteboook lid microphones capture level. It can be improved with enable of DRC and beamformer TDFB components in topologies build. Since the same pipelines are not used for other microphone endpoints, it is safe to apply fixed processing configuration. The topologies with -2ch are set with DRC profile "dmic_default" and with TDFB profile "line2_generic_pm10deg". It provides about +10 dB boost to sound level and a user facing microphone stereo beam that improves signal-to-noise ratio. The user space controls are disabled by default, so the user needs to set the features on with alsamixer or amixer conmmands. The 4ch topologies are set with passthrough blob for beamformer. With a lot of variation in array form factors there is no generic setting available. The DRC setting is the same for 4ch and provides +10 dB boost. The impacted topologies are: - sof-mtl-rt711-4ch - sof-mtl-rt711-l0-rt1316-l3-4ch - sof-mtl-rt711-l0-rt1316-l3-2ch - sof-mtl-rt712-l0-4ch - sof-mtl-rt712-l0-2ch Signed-off-by: Seppo Ingalsuo <[email protected]>
The topologies for PCs those have PCH DMIC enabled have too silent noteboook lid microphones capture level. It can be improved with enable of DRC and beamformer TDFB components in topologies build. Since the same pipelines are not used for other microphone endpoints, it is safe to apply fixed processing configuration. The topologies with -2ch are set with DRC profile "dmic_default" and with TDFB profile "line2_generic_pm10deg". It provides about +10 dB boost to sound level and a user facing microphone stereo beam that improves signal-to-noise ratio. The user space controls are disabled by default, so the user needs to set the features on with alsamixer or amixer conmmands. The 4ch topologies are set with passthrough blob for beamformer. With a lot of variation in array form factors there is no generic setting available. The DRC setting is the same for 4ch and provides +10 dB boost. The impacted topologies are: - sof-lnl-rt711-4ch - sof-lnl-rt713-l0-rt1318-l1-2ch - sof-lnl-cs42l43-l0-4ch - sof-lnl-cs42l43-l0-cs35l56-l3-2ch Signed-off-by: Seppo Ingalsuo <[email protected]>
The topologies for PCs those have PCH DMIC enabled have too silent noteboook lid microphones capture level. It can be improved with enable of DRC and beamformer TDFB components in topologies build. Since the same pipelines are not used for other microphone endpoints, it is safe to apply fixed processing configuration. The topologies with -2ch are set with DRC profile "dmic_default" and with TDFB profile "line2_generic_pm10deg". It provides about +10 dB boost to sound level and a user facing microphone stereo beam that improves signal-to-noise ratio. The user space controls are disabled by default, so the user needs to set the features on with alsamixer or amixer conmmands. The 4ch topologies are set with passthrough blob for beamformer. With a lot of variation in array form factors there is no generic setting available. The DRC setting is the same for 4ch and provides +10 dB boost. The impacted topologies are: - sof-ptl-rt722-4ch Signed-off-by: Seppo Ingalsuo <[email protected]>
|
This should fix issue #9681 |
|
The change adds the beamformer (TDFB) and dynamic range control (DRC) to PCH DMIC capture host pipeline:
|
|
SOFCI TEST |
The existing default is high-pass 40 Hz with 0 dB gain. This change amplifies captured sound from DMIC by 20 dB. The same amplifying IIR type is used in hda-generic. The change is for now, but not the final solution. After we have UCM2 control for capture processing in SOF we can shift applying more gain to the DRC component that will provide more robustness with silent and loud environments. Signed-off-by: Seppo Ingalsuo <[email protected]>
|
Added 3rd commit, the comment and the new example recording from #9681 indicates that more amplification is needed. |
There was a problem hiding this comment.
@singalsu , do we need to enable nhlt-sof-ptl-rt721-4ch.bin as well?
There was a problem hiding this comment.
I was wondering what it is due to comment "RT721 eval board with PCH-DMIC" so I did not change it. Is it OK to add the processing to it too?
I can do a separate PR to address it. This PR should be merged quick to main branch to be able to make the v2.11-stable update for the IIR gain.
There was a problem hiding this comment.
@bardliao ,I assume this would also be a customer board configuration for future use. What are your thoughts?
There was a problem hiding this comment.
I was wondering if it's a topology for a Chromebook or a PC. In this step I'm adding processing to PCs running Linux. With Chromebooks similar type processing has been on the user space and/or in the closed firmware component, so it has not been enabled.
There was a problem hiding this comment.
I think the sof-ptl-rt721-4ch topology is for chrome EVB.
There was a problem hiding this comment.
Thanks! OK, I'll leave it without this change for now.
|
SOFCI TEST |
singalsu
changed the title
|
I have in every commit typo "PCM DMIC" when it should be "PCH DMIC", I think best to change to avoid confusion. Jenkins test results: https://sof-ci.01.org/sofpr/PR9690/build9622/devicetest/index.html https://sof-ci.01.org/sofpr/PR9690/build9623/devicetest/index.html https://sof-ci.01.org/sofpr/PR9690/build9621/devicetest/index.html https://sof-ci.01.org/sof-pr-viewer/#/build/PR9690/build14364789 |
|
sof-docs fail and Intel LNL fails all known and tracked in https://github.com/thesofproject/sof/issues?q=is%3Aissue+is%3Aopen+label%3A%22Known+PR+Failures%22+ @singalsu @macchian let's handle rt721 case separately in a follow-up PR |
No description provided.