Audio Processing App

Upload audio files (MP3 or M4A) for processing and transcription.
Intended use case: transcription of interviews.

Silence Removal Settings

Default settings are working very well. Silence removal helps to reduce hallucination.

Chunking Settings

Chunking reduces the load on the model. 10min chunks work really good.

Transcription Settings

tiny is the fastest, but the worst quality. Large-v3-turbo is the best, but slower.

Whisper Model Size
Language