noScribe
Local AI transcription app for research interviews and journalism. Runs on Windows, macOS and Linux, with speaker identification and a transcript editor.
noScribe is a desktop app for researchers and journalists who need transcripts of recorded interviews without sending confidential audio to a cloud service. It processes recordings entirely on your computer and can distinguish between speakers, helping you keep track of who said what in a conversation.
The app combines OpenAI's Whisper speech recognition with faster-whisper and pyannote for speaker identification. Its graphical interface brings these tools together for interview transcription, and its multilingual recognition supports interviews in different languages.
A built-in editor lets you review, verify and correct the generated transcript. That makes human review part of the same application, useful when the wording of an interview matters to a research finding or a quotation. Speaker identification and transcript correction are both relevant to its focus on qualitative social research and journalism.
noScribe is free and open source under the GNU General Public License v3.0 (GPL-3.0). It runs on Windows, macOS and Linux. Audio processing stays on the user's computer rather than going to a hosted transcription service.
