Thoth

Thoth is a transcription and meeting assistant tool for macOS, focusing on data privacy by processing audio locally. It offers optional cloud AI features that work with your chosen provider using your API keys. Thoth makes recording system audio on macOS easy, allowing simultaneous recording of microphone and system audio and compatibility with platforms like Zoom, Teams, Google Meet, and Webex. Users can isolate audio, import existing audio files, and record up to 30 minutes (mic) or 15 minutes (system/mixed) per session for free.

Thoth has an offline transcription feature for confidentiality, offers five WhisperKit model sizes, 99 languages with auto-detection, and live transcription with three engine options. It uses Pyannote v4 models for neural speaker diarization, identifying up to 8 speakers accurately. Users can add custom vocabulary, integrated into the decoder. The free tier includes Base and Small models, the Pro version unlocks the Large V3 Turbo model.

Thoth provides features to clean up, summarize, translate, and extract insights either locally or via your cloud provider. It enhances transcripts, refines speaker attribution, provides summaries, and offers translation. Results are saved for later review. Free users get 10 AI enhancements per month, Pro users have unlimited enhancements.