AudioToText
A local audio transcription tool that uses Whisper, exports subtitles and supports cloud translation through DeepL on macOS, Windows and Linux.
AudioToText turns recordings into text and captions for people who need transcripts or subtitles in different languages. It can process audio on your computer with Whisper or use cloud processing.
- Multilingual transcription: Whisper detects the spoken language and produces a transcript in that language, with support for processing multiple audio files.
- Translation: Whisper translates speech into English; translation into other supported languages uses DeepL's cloud API and requires a DeepL Developer account and API key.
- Transcript and subtitle exports: Save readable TXT transcripts, VTT or SRT captions, and TSV or JSON files for further processing.
- Long recordings: The tool can divide audio at pauses to reduce memory use during local transcription.
The Python command-line tool and Jupyter notebook run locally on macOS, Windows and Linux. Local Whisper processing uses your hardware, with GPU acceleration or slower CPU execution. Smaller models such as tiny, base and small are available for less powerful computers.
The optional OpenAI API sends audio to the cloud for transcription. Google Colab can also run the notebook on a cloud GPU. DeepL receives transcript text for translation and can share context between lines or adjust formality in supported languages. If DeepL doesn't support the original language, Whisper can first translate the speech into English for DeepL to translate into the target language.
