AudioNotes

Why we included this project

AudioNotes runs entirely on your own computer, so turning recordings into clean notes doesn't rely on cloud AI services. FunASR handles speech recognition locally and an Ollama-hosted model organizes the transcript into structured Markdown, which keeps recordings, transcripts, and chat history on your machine. That local-first setup is the main draw if you work with confidential audio that can't be sent to third-party APIs. Beyond transcription, you can ask follow-up questions about the material and reopen earlier notes from a saved history. The default model pair is picked for solid Chinese recognition on ordinary hardware, with an option to switch to a multilingual model for recordings in other languages.