How transcription works
Choose where the listening happens: on this computer, with a cloud provider, or on another computer on your network.
Choose a mode
Set Mode in Settings → Transcription. A change made while transcription is running applies the next time it starts.
- Local, the default: a speech model on this computer listens. No audio leaves the computer, and no internet is needed once the model is installed.
- Cloud: the audio goes to a provider you choose, which sends back the words.
- Remote: the words come from a transcription server on your network, such as another AutoSermon. This computer’s microphone is not used.
Cloud providers
Pick a Provider and paste its API key: OpenAI, Groq or Google Gemini. The key is the same one Settings → AI providers uses. The provider bills you for what you use, on any AutoSermon plan.
Remote: another AutoSermon
Another computer running AutoSermon does the listening. This computer keeps its own transcript, verse detection and history.
- On the computer that listens, press Start server in Settings → Network. Turn on Start with AutoSermon to keep it on.
- On this computer, set Mode to Remote.
- In Server address, enter the other computer’s address and port as its Settings → Network shows them, for example 192.168.1.20:9876. Without a port, 9876 is used.
- Press Test. Reachable means that computer answered; Unreachable says what to turn on.
Any server that streams plain text, or JSON with a text field, over a WebSocket also works: enter its full URL, path included.
Recognition language
Set the language of the sermon on the same page: English (US), English (UK), Spanish, French, German, Portuguese, Italian, Dutch, Indonesian, Korean or Chinese.
NVIDIA Nemotron + Parakeet hears English only. For any other language, Local uses Small or Base. If neither is installed, the page names the one to install.
Microphone
The Microphone tab on Settings → Transcription sets the input, Noise reduction (off by default) and Sample rate (48 kHz by default), and shows the Input level when you press Test. See Audio setup.
If a mode cannot start
When Local or Cloud cannot start for a reason the other can solve, such as a model not installed or no API key, AutoSermon uses the other and says so. Remote stops and gives the reason. If transcription cannot start at all, the Live transcription panel says why, with Try again and Transcription settings.
Using Cloud in place of Local sends the audio to your cloud provider. If it must stay on this computer, install the speech model before the service.