Describe the bug
Transcription from mic always fails with Error: Failed to transcribe chunk 0: File transcription not supported. Use microphone recording with transcribe_raw().
File transcription seems to work fine, however. See the logs below.
According to the verbose logs audio seems to be recorded, so I'm confused where the problem occurs exactly.
Environment
- OS: Arch Linux
- Whis version: 0.7.2
- Installation method: cargo install
Additional context
whis -v --print -f ~/test.wav
[verbose] Loading whisper model from: /home/user/.local/share/whis/models/ggml-small.bin
[verbose] Whisper model loaded successfully
[verbose] Unloading whisper engine from cache
Try a recording using FFmpeg and my laptop microphone.
whis -v --print
[verbose] Audio device: Default Audio Device
[verbose] Preloading whisper model in background...
[verbose] Loading whisper model from: /home/user/.local/share/whis/models/ggml-small.bin
[verbose] Audio device: 44100 Hz, 2 channel(s) -> resampling to 16000 Hz mono
[verbose] VAD enabled (threshold: 0.50)
[verbose] Whisper model loaded successfully
[verbose] Created final chunk 0 (9.9s)
[verbose] Recorded 157696 samples (9.9s at 16000 Hz mono)
Error: Failed to transcribe chunk 0: File transcription not supported. Use microphone recording with transcribe_raw().
Describe the bug
Transcription from mic always fails with
Error: Failed to transcribe chunk 0: File transcription not supported. Use microphone recording with transcribe_raw().File transcription seems to work fine, however. See the logs below.
According to the verbose logs audio seems to be recorded, so I'm confused where the problem occurs exactly.
Environment
Additional context