Feb. 17, 2026
8 a.m.
I still think what I mentioned elsewhere and in my first comment here. In particular, notice that we don't want to match native WASAPI's resampler in our dsound: WASAPI's is indeed higher quality, and much, much slower, than the native dsound one. For "mmdevapi" we currently just use pipewire / pulseaudio / coreaudio's sample rate conversion and that seems like a reasonable fit. Both pipewire and pulseaudio allow tweaking the resample quality as well. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10056#note_129800