Playback speed adjustment and silence skipping (and possibly custom audio processing injected by the application) are currently disabled by DefaultAudioSink if the input uses ENCODING_PCM_FLOAT.
I think the idea of this is to avoid resampling 32-bit floating point into into 16-bit integer precision PCM. However, in cases where we actually want to do processing that changes the audio, such as playback speed adjustment and silence skipping, then that should take precedence and we should resample to 16-bit integer precision to enable that processing to take place.
Playback speed adjustment and silence skipping (and possibly custom audio processing injected by the application) are currently disabled by
DefaultAudioSinkif the input usesENCODING_PCM_FLOAT.I think the idea of this is to avoid resampling 32-bit floating point into into 16-bit integer precision PCM. However, in cases where we actually want to do processing that changes the audio, such as playback speed adjustment and silence skipping, then that should take precedence and we should resample to 16-bit integer precision to enable that processing to take place.