Meta for macOS

Availability

Meta 1.0.0 is available as a TypeWhisper for macOS plugin release. Its current source code is maintained on the repository’s main branch.

What the macOS add-on does

The Meta add-on combines two cloud capabilities:

• Muse Voice Transcribe handles live dictation and uploaded audio files. It can return timestamps and, when enabled, speaker labels.
• Muse Spark serves as a prompt model for TypeWhisper workflows.

The add-on supports language hints and dictionary terms for both live and uploaded transcription. Translation is not supported.

Settings

Setting	What it controls
Meta Model API	Enter the API key, reveal or hide it, save it in the macOS Keychain, and refresh the available models. Remove deletes the stored credential.
Transcription Model	Selects the Muse Voice Transcribe model. Muse Voice Transcribe 1.0 is the built-in fallback.
Speaker diarization	Switches Meta from push-to-talk mode to diarization mode. Speaker labels apply to both real-time and uploaded audio.
Prompt Model	Selects the Muse Spark model used for workflow prompt processing.

The built-in prompt-model catalog contains Muse Spark 1.2, Muse Spark 1.1, and Muse Spark 1.2 Contributor. Contributor models display an explicit warning that prompts and completions may be used to train future Meta models.

Setup tutorial

1. Create an API key in the Meta Model API developer portal.
2. Open TypeWhisper > Settings > Integrations.
3. Find Meta under Discover, install it, and make sure it is enabled.
4. Open the add-on settings and paste the API key.
5. Select Save and Refresh Models. A successful request shows Connected and stores the key in the macOS Keychain.
6. Choose a Transcription Model and a Prompt Model.
7. Enable Speaker diarization only for recordings where speaker-labelled output is useful.

Use Muse Voice Transcribe

1. Open or create a TypeWhisper profile and select Meta as its transcription engine.
2. Choose the expected language or languages. Supported selections are sent to Meta as language bias.
3. Add names and specialist vocabulary to the TypeWhisper dictionary. Supported terms are forwarded as keywords in both live and file transcription.
4. For interviews or meetings, enable Speaker diarization before recording or importing the file.
5. Run a short test. In diarization mode, completed turns are formatted with labels such as Speaker 1: when Meta returns them.

Real-time transcription streams 16 kHz PCM through Meta’s WebSocket API. When recording stops, TypeWhisper waits for Meta’s final event before completing the transcript.

Use Muse Spark in a workflow

1. Create or edit a TypeWhisper workflow with prompt processing.
2. Select Meta as the prompt provider.
3. Choose the Muse Spark model configured in the add-on settings.
4. Test the workflow on representative dictated text before applying it automatically.
5. Avoid a Contributor model for sensitive content unless its displayed training notice fits your use case.

Troubleshooting

• Connection Failed: Check the API key and Meta API availability, then select Save and Refresh Models again.
• A model is missing: Refresh the catalog. Successfully loaded models are cached so the settings can still show them offline.
• No speaker labels appear: Confirm that Speaker diarization was enabled before the live session or file transcription and that Meta returned speaker information.
• Translation fails: Muse Voice Transcribe does not expose translation through this add-on.

Cloud processing and privacy

Audio is sent to Meta for Muse Voice Transcribe, and workflow text is sent to Meta for Muse Spark. Prompt-processing requests set store to false, but Contributor models still show Meta’s training warning. Review Meta’s current terms, privacy information, regional availability, and pricing before using sensitive content.