New reworked driver for Wolfson/Cirrus Logic audio card

Can I boost DMIC sensitivity?

Can I boost DMIC sensitivity?

Hi Andrey!

Can I boost DMIC sensitivity?

Yes. You can change "IN2L Digital Volume" and "IN2R Digital Volume". By default these are set to 116 which is -6dB. Setting these to 128 would be 0dB (each step is 0.5dB). You can also set them to more than 128 - just try different values and see what works best.

BTW: The purpose of the "usecase" mixer scripts is to give you a quick start, usually you'll have to tweak the settings (boost/attenuate signals, add/remove filters, use different routing etc). Remembering which input/output corresponds to which mixer can be hard, therefore I created a bunch of helper functions and definitions in rpi-cirrus-functions.sh to simplify that. Feel free to change that to your needs, modify it or use it in your own scripts.

For example, the "Record_from_DMIC.sh" script calls the "record_from_dmic" function in rpi-cirrus-functions.sh. It looks like that:

# args: [ VOLUME [ DIGITAL_VOLUME ] ]
record_from_dmic() {
        reset_dmic_in
        reset_filter
        reset_rpi_in

        # default input gain 0dB and digital gain -6dB
        setup_dmic_in "${1:-0}" "${2:-116}"

        # route input through high pass filter to remove DC
        setup_high_pass_filter $dmic_in_signals
        set_mixer $rpi_in_signals $filter_signals
}

You can pass 2 additional arguments to that function, volume (which should be 0 when using DMIC) and the digital volume (eg 128). If these are not set, the default values 0 and 116 will be used when calling setup_dmic_in.

If you want you can also remove the high-pass filter, drop the "set_high_pass_filter" line and change the set_mixer call below to "set_mixer $rpi_in_signals $dmic_in_signals" - then DMIC won't be routed through the high-pass filter but sent directly to the RPi audio input.

If you want to mix multiple (analog) inputs into your output signal have a look at what "Cirrus_listen.sh" does, eg in the case of listening to DMIC on line-out:

First it configures the DMIC input, like the record function:

setup_dmic_in 0 116

Then it sets the lineout volume to the default (128 means 0dB)

mixer "${line_out} Digital Volume" 128

and routes the input signal (in this case "$dmic_in_signals) to the specified input of the line-out mixer (default is 2):

set_mixer $line_out_signals $dmic_in_signals 32 2

Each mixer block inside the WM5102 has 4 inputs, the "Playback" scripts/functions use the first one, the listen scripts by default the second one, so if you like you can add 2 more signals (eg line-in and headset-in) to the remaining 2 mixer inputs.

The "32" in the set_mixer call specifies the volume of the input. 32 is 0dB. Change it to your needs, for example to boost dmic-in or attenuate line-in.

so long,

Hias

Hi , i am looking  for a way to combine Linein and DMIC. This worked for me in the old use-case script files but I do not see a way to do this with the new use-case-ng scripts. Could you shine your light on this?

Hi , i am looking for a way to combine Linein and DMIC. This worked for me in the old use-case script files but I do not see a way to do this with the new use-case-ng scripts. Could you shine your light on this?

You can either to this with the helper functions in rpi-cirrus-functions.sh or manually via amixer commands (like before, just change the soundcard name).

If you like to test with the helper functions you can simply source the helper file in your current shell and then also use the helper functions on the command line. For example

pi@rpi3:~ $ . bin/rpi-cirrus-functions.sh
pi@rpi3:~ $ record_from_linein
pi@rpi3:~ $ setup_dmic_in 0 116
pi@rpi3:~ $ set_mixer $filter_signals $dmic_in_signals 32 2

This'll first configure for normal linein recording where the line input is sent through a high-pass filter.

Then setup the dmic input: 0 is the analog gain (keep this at 0 for dmic), 116 is the digital gain, -6dB.

In the third line the DMIC input is mixed together with the line input in the highpass filter. LineIn uses the default first input of the filter, dmic the second input - this is the 2 at the end. The 32 before that is the input gain at the filter which means 0dB.

Another possibility would be to drop the filter and mix both signals together in the digital output mixer to the RPi:

pi@rpi3:~ $ record_from_linein
pi@rpi3:~ $ setup_dmic_in 0 116
pi@rpi3:~ $ set_mixer $rpi_in_signals $line_in_signals 32 1
pi@rpi3:~ $ set_mixer $rpi_in_signals $dmic_in_signals 32 2

BTW: if you enter "set -x" you can see the actual mixer commands and helper functions being called.

so long,

Hias

Hello Matthias,

thx for your response, I am using the helper function file as you stated in your response with good results!

i was also looking for the DMIC on/off switch but soon realized I could as well set DMIC to 0,0 when not using the microphone!

thanks for your answer!