Description
At the moment, the library records audio only in .m4a (AAC).
While this is fine for playback, it limits use cases where raw PCM audio is required.
A common example is on-device speech-to-text and ML models (e.g. Whisper, Vosk, DeepSpeech), which expect uncompressed PCM input such as mono 16-bit WAV. When using the current .m4a output, audio must be transcoded before processing, adding extra complexity, CPU usage, and latency on mobile.
Why this would be useful
- Enables direct compatibility with STT / ML pipelines that require PCM audio
- Avoids post-processing or FFmpeg-based transcoding
- Allows lossless audio capture for analysis and AI use cases
- Makes the library more flexible for non-playback scenarios
Suggested improvement
Expose a recording option to choose the audio format / encoder, for example:
- aac (current behavior)
- wav (uncompressed PCM)
The library could then:
- Record PCM audio natively (.wav) on both iOS and Android
- Automatically select the correct file extension based on the chosen format
Thanks in advance!
Description
At the moment, the library records audio only in .m4a (AAC).
While this is fine for playback, it limits use cases where raw PCM audio is required.
A common example is on-device speech-to-text and ML models (e.g. Whisper, Vosk, DeepSpeech), which expect uncompressed PCM input such as mono 16-bit WAV. When using the current .m4a output, audio must be transcoded before processing, adding extra complexity, CPU usage, and latency on mobile.
Why this would be useful
Suggested improvement
Expose a recording option to choose the audio format / encoder, for example:
The library could then:
Thanks in advance!