[BUG] ramp time extended by 0.03s after capture gain curve_duration change

[keqiaozhang]

Describe the bug
Observed this issue on all IPC4-nocodec(single core) platforms. it caused alsabat test failures.
The ramp time extended by 0.03s after capture gain curve_duration change(d0d74a4):

To Reproduce
check the alsabat wav file

Reproduction Rate
100%

Environment

1. Branch name and commit hash of the 2 repositories: sof (firmware/topology) and linux (kernel driver).
   • Kernel: {cffcac365eac}
   • SOF: {capture gain curve_duration change}
2. Name of the topology file
   • Topology: {nocodec tplg}
3. Name of the platform(s) on which the bug is observed.
   • Platform: {NOCODEC-IPC4 platforms}

[marc-hb]

The commit message says "This change increases the ramp duration from 20 ms to 50 ms"

So it's not a bug, it's a feature? :-)

[keqiaozhang]

So it's not a bug, it's a feature? :-)

Not sure about this, but alsabat cannot survive from such long ramp time.

[singalsu]

Ouch, but there's justification for this change. 1) capture artefacts mitigation accidentally dropped from tplg2 vs. tplg1 2) we can avoid a big MCPS peak with this lower rate ramp.

Is there a way to ask alsabat to ignore a bit longer audio from beginning?

[singalsu]

If not possible to change alsabat we can have a test topology with short ramp?