Audio playback stuttering problem

[NeilTohno]

Important Information

Provide following Information:

• Windows Version
  Windows 10 22H2
• Source of the mpv binary
  https://github.com/shinchiro/mpv-winbuild-cmake/releases/tag/20240426

I added the following code to configuration file, will this cause audio playback issues?
Thanks.

# Audio #

audio-buffer=0.6
ad-lavc-threads=12
ao=wasapi
audio-exclusive=yes
wasapi-exclusive-buffer=6000 #min

audio-format=float
audio-channels=stereo
audio-swresample-o=resampler=soxr,precision=33
audio-swresample-o-append=cheby=1
audio-swresample-o-append=cutoff=1.00
audio-swresample-o-append=osf=fltp

# fix bgm too loud and speech too quiet
#af=lavfi="loudnorm=I=-16:TP=-3:LRA=4"
af=lavfi="loudnorm=I=-17:TP=-1:LRA=11"

#13996 (comment)

Log file:

[ 9.881][v][ao/wasapi] Accepted as stereo s16 @ 48000hz -> stereo s16 (16/16 bits) @ 48000hz (exclusive)
[ 9.881][d][ao/wasapi] Fixing format
[ 9.881][d][ao/wasapi] IAudioClient::GetDevicePeriod
[ 9.881][v][ao/wasapi] Device period: default 10000 us, minimum 3000 us
[ 9.881][v][ao/wasapi] Requested buffer duration: 6000 us
[ 9.881][v][ao/wasapi] Trying buffer duration 6000 us
[ 9.881][d][ao/wasapi] IAudioClient::Initialize
[ 9.933][d][ao/wasapi] IAudioClient::Initialize pRenderClient
[ 9.933][d][ao/wasapi] IAudioClient::Initialize IAudioClient_SetEventHandle
[ 9.933][d][ao/wasapi] IAudioClient::Initialize IAudioClient_GetBufferSize
[ 9.933][v][ao/wasapi] Buffer frame count: 288 (6000 us)
[ 9.933][v][ao/wasapi] IAudioClock::GetFrequency gave a frequency of 48000.
[ 9.934][d][ao/wasapi] Activating pEndpointVolume interface
[ 9.935][d][ao/wasapi] IAudioEndpointVolume::QueryHardwareSupport
[ 9.935][d][ao/wasapi] Entering dispatch loop
[ 9.935][d][ao/wasapi] Init wasapi done
[ 9.935][v][ao/wasapi] device buffer: 288 samples.
[ 9.935][v][ao/wasapi] using soft-buffer of 28800 samples.
[ 9.935][i][cplayer] AO: [wasapi] 48000Hz stereo 2ch s16
[ 9.935][v][cplayer] AO: Description: Windows WASAPI audio output (event mode)
[ 9.935][v][autoconvert] inserting resampler
[ 9.935][v][swresample] format change, reinitializing resampler
[ 9.936][v][swresample] 192000Hz stereo double -> 48000Hz stereo s16
[ 9.948][v][af] [out] 48000Hz stereo 2ch s16

I found lost of notification,

mpv.com
ao/wasapi: Under-run: Device delay: -1400 ns
ao/wasapi: Under-run: Device delay: -1800 ns
ao/wasapi: Under-run: Device delay: -1400 ns
ao/wasapi: Under-run: Device delay: -1300 ns
ao/wasapi: Under-run: Device delay: -1300 ns
ao/wasapi: Under-run: Device delay: -1700 ns
ao/wasapi: Under-run: Device delay: -1800 ns
ao/wasapi: Under-run: Device delay: -1700 ns

[ruihe774]

May I ask: what's the purpose of your settings? What is your goal? Why you have to touch the default settings?

FYI, as a normal user with normal headphones, I just use the default audio settings in mpv.

[NeilTohno]

fix bgm too loud and speech too quiet
af=lavfi="loudnorm=I=-17:TP=-1:LRA=11"

Usually, I add this code in the mpv‘s configuration file, I concern about the quality of audio resampling, yeah.

[ruihe774]

I concern about the quality of audio resampling

Please do an ABX test to convince yourself that a customized config is indeed better. Nothing is "improved" if you cannot notice a difference.

fix bgm too loud and speech too quiet

Try --audio-normalize-downmix first.

[NeilTohno]

48 kHz - 192 kHz - 48 kHz
Do twice resampling, I can say, it's a real problem.

--audio-normalize-downmix

This code is useful for converting 5.1 channels to 2.0 channels.

[ruihe774]

This code is useful for converting 5.1 channels to 2.0 channels.

I misunderstood your use case. I thought you were downmixing.

Do twice resampling, I can say, it's a real problem.

No, IMO.

af=lavfi="loudnorm=I=-17:TP=-1:LRA=11"

The problem you want to solve, IIRC, is exactly replaygain aims to solve.

BTW, from ffmpeg's documentation:

However, it is recommended to run the normalization with two passes, extracting the measured values from the first run, then using the values in a second run with linear normalization enabled. See the loudnorm filter documentation for more.

I believe it is not suitable to be used in playback.

[NeilTohno]

I believe it is not suitable to be used in playback.

There is no better method than this, so one pass is ok.

[NeilTohno]

I will test those codes with following option, funny.

audio-resample-max-output-size=100
msg-level=all=status,ao/wasapi=trace

mpv_f.zip

[NeilTohno]

This should be ok, but problem happens again.

I'm experiencing audio popping and lagging issues when playing video files. Music files play back without any problems.

interpolation=yes
interpolation-threshold=-1
# autosync=30
framedrop=vo
video-sync=display-vdrop