Core workflows
Synthezia creates a session from selected media or a recording, normalizes the audio, and transcribes it before you work with a summary or questions about that session.
Current workflow building blocks
Section titled “Current workflow building blocks”| Capability | Status | Current behavior |
|---|---|---|
| Imported audio and video processing | Available | Select an audio or video file, create a session, convert the selected media, transcribe it, and start summary generation. |
| Live microphone recording | Beta | Create a session while capturing the default microphone and then run the normal conversion, transcription, and summary pipeline. |
| Local Whisper transcription | Available | Transcribe normalized audio by running a local whisper.cpp command-line binary with a locally stored model. |
| Local Ollama summaries and chat | Available | Generate summaries, session chat answers, and global chat answers through a local Ollama HTTP runtime when Local Mode is selected. |
| Per-session chat | Available | Ask questions using a session transcript, optional summary, prior chat turns, and the selected local or remote AI provider. |
Processing choices
Section titled “Processing choices”Local Mode processes transcripts with local models after the required models are installed. Selecting an OpenAI-compatible provider changes the data boundary for the flows that use it; review that boundary before sending session content.
Continue in AI and models or review privacy and security.