On-device and open source Whisper dictation (github.com)

Local Flow has launched an open-source, on-device dictation tool powered by Whisper, designed for macOS users. This innovative application prioritizes privacy by ensuring that all voice transcription occurs locally on the user's machine without sending data to external servers. Users can select from various Whisper models—Best, Classic, or Tiny—and customize functionalities such as microphone selection and transcription shortcuts in the app's settings. The tool features a live waveform display during dictation and is capable of processing long audio recordings by splitting them into manageable segments for improved transcription accuracy. The significance of Local Flow lies in its blend of advanced transcription technology with robust privacy measures, catering to the growing demand for secure and efficient dictation solutions in the AI/ML community. By allowing cloud integration with providers like OpenAI and ElevenLabs, it offers flexibility while still maintaining stringent data protection standards. Additionally, its self-updating feature and customizable cleanup rules enhance its usability, making it a powerful tool for professionals across various fields looking for seamless voice-to-text capabilities without compromising their information security.