Comment by xuhu

For one, audio transcription services that use Whisper will sample the input down to 16Khz mono first.