| Local Mode |
The default path that uses local Whisper and the bundled local Ollama runtime after required models are available. |
| External API |
The optional OpenAI-compatible provider mode. It changes the boundary for generation and provider-aware embeddings. |
| Whisper model |
A local whisper.cpp model file used to transcribe audio. |
| Ollama model |
A local model used by the bundled Ollama runtime for summaries and chat. |
| Embedding |
A vector representation used to retrieve related session or transcript content for search and chat. |
| Session |
A local record for imported media or a live recording, including processing state and derived content. |
| Transcript |
The text returned by local Whisper from normalized audio. |
| Summary mode |
The instruction structure used to organize generated summary output. |
| Global index |
Local text and vector search data built from session content for cross-session retrieval. |
| Loopback |
A network address that points back to the same device; Synthezia uses it for its local Ollama runtime. |
| Model requirement |
The dialog shown when an action needs a local model that is not installed. |
