Software · Own product · 2026 · In development
AI Transcription Studio
A desktop app that transcribes audio and video on your own machine, tells speakers apart, and exports subtitles and transcripts, with a batch queue for whole folders.
- 3
- transcription engines to choose from
- 4
- export formats, subtitles included
- Local
- processing: audio never leaves the machine
A Danube Labs product in development. It is closed source and not released yet; this page presents what it does.
Interviews, meetings and lectures are often too sensitive to upload anywhere. AI Transcription Studio runs every step locally on your GPU or CPU, so the audio never leaves the computer.
What it does
- Selectable engines: Faster-Whisper for speed, WhisperX for precise word timing, OpenAI Whisper as a fallback
- Speaker diarization with Pyannote, so the transcript says who spoke when
- Optional noise reduction and automatic conversion of video and audio formats through FFmpeg
- A job queue that takes whole folders, pauses and resumes, skips files already done, and keeps going when one file fails
- Exports to TXT, JSON, SRT and VTT
- Model, device and quantisation settings, with GPU memory released between transcription and diarization
The output in the screenshot is blurred because it is a real interview.