sandrohanea / whisper.net

Whisper.net. Speech to text made simple using Whisper Models
MIT License
547 stars 84 forks source link

Not directly supporting MP3 format? #156

Closed dfengpo closed 8 months ago

dfengpo commented 8 months ago

Does it support recording in WAV format by default? Whisper itself supports all recording formats, why doesn't Whisper.net support all recording formats? Otherwise, we need to convert the format separately

sandrohanea commented 8 months ago

No other format than wave is supported natively by whisper.

The python version (OpenAI) of the whisper is just using ffmpeg to decode other formats, but we don't want to add a hard-dependency on ffmpeg as in dotnet, most of the people might want to use NAudio.

Also discussed in https://github.com/sandrohanea/whisper.net/issues/27

dfengpo commented 8 months ago

However, NAudio does not support Linux as it integrates too many Windows related APIs