Meta for Windows

Availability

Meta 1.2.14 is available in Settings > Discover plugins in TypeWhisper for Windows 1.1.0 or newer and as a plugin release. Its current source code is maintained on the repository’s main branch.

What the Windows add-on does

Version 1.2.14 provides two capabilities through one Meta API key:

• Transcription: Muse Voice Transcribe handles live dictation, finished recordings and audio files. It can label speakers when Speaker labels is on.
• LLM workflows: Muse Spark processes TypeWhisper workflow prompts. The built-in catalog contains Muse Spark 1.2 and 1.1.

Both capabilities use the same API key, which TypeWhisper keeps in encrypted Windows user storage. Refresh models loads the compatible models available to your key from Meta’s /v1/models endpoint. If a refresh fails, the previous catalog remains.

Settings

The add-on’s page in the Settings sidebar shows these options. Changes to Speaker labels, Text model and Reasoning effort take effect when you select Save settings.

Setting	What it controls
API key	Stores the Meta Model API key. Enter a new key to replace the saved one; Remove saved key deletes it.
Check connection	Checks the saved key against Meta. No audio is uploaded, and the model catalog stays unchanged.
Refresh models	Loads the muse-voice-transcribe-* and muse-spark-* models available to your key and reports how many transcription and text models were found.
Transcription model	Choosing a Muse Voice Transcribe model makes it the model for dictation. Before the first refresh, Muse Voice Transcribe 1.0 is available. While the shown model is not in use, Use selected model appears and switches dictation to it.
Speaker labels	Off or On. When on, Meta identifies speakers in live and recorded transcription. Each speaker turn starts on its own line with the speaker label that Meta returns.
Text model	The default Muse Spark model for text processing. Before the first refresh, Muse Spark 1.2 and Muse Spark 1.1 are available.
Reasoning effort	Minimal, Low, Medium, High or Very high. Higher effort can improve complex tasks but increases response time. The default is Medium.

Changing the API key clears the loaded model catalog. Model selections that the built-in catalog does not contain return to Muse Voice Transcribe 1.0 and Muse Spark 1.2. Select Refresh models to load the catalog for the new key.

Setup tutorial

1. Make sure you use TypeWhisper for Windows 1.1.0 or newer, and create an API key in the Meta Model API developer portal.
2. In TypeWhisper, open Settings > Discover plugins and install Meta. The add-on then appears in the Settings sidebar, and its page opens.
3. Paste the key into API key and select Save settings.
4. Select Check connection, then Refresh models. The page reports how many transcription and text models were loaded.
5. Under Transcription model, choose a Muse Voice Transcribe model and, if it appears, select Use selected model.
6. Set Speaker labels to On only when speaker-labelled text is useful.
7. Choose a Text model and leave Reasoning effort at Medium for a balanced start. Then select Save settings.

Use Muse Voice Transcribe

1. Check that the add-on’s page shows Selected for dictation. under Transcription model.
2. Under Settings > Dictation, set the Spoken language, or leave it on Automatic and choose Preferred languages. The add-on passes supported languages to Meta as language bias.
3. Add important names and specialist terms to the Dictionary. The add-on sends up to 100 of them, with 600 characters in total, to Meta as keywords. This works for live and recorded audio and does not disable streaming.
4. Turn on Live transcription under Settings > Appearance if you want Meta to transcribe while you speak. Without it, the add-on uploads the finished recording.
5. Record a short test and check the final text before using Meta for longer audio.

In a live session with speaker labels on, the add-on waits until Meta has delivered all completed speaker turns before it finishes the result. If the live connection is interrupted, TypeWhisper transcribes the full recording after you stop. Translation is not supported.

Use Muse Spark in a workflow

1. Create or edit a workflow under Workflows.
2. Choose Meta under Provider and a Muse Spark model under Model.
3. Start with Medium reasoning effort and test the workflow on representative dictated text.
4. Increase Reasoning effort on the add-on’s page for more involved instructions, or reduce it when faster, simpler processing is sufficient. The level applies to all Muse Spark requests from the add-on.

Troubleshooting

• The transcription model cannot be selected: The page asks you to complete the provider configuration or says that an API key is required. Save an API key first.
• Check connection fails: Check the key in Meta’s developer portal, paste it again into API key, select Save settings and check the connection again.
• Refreshing models fails: The previous catalog remains, so the saved or built-in models stay usable. Check the connection and Meta API availability, then select Refresh models again.
• No speaker labels appear: Confirm that Speaker labels was set to On and saved before the recording or file transcription, and that Meta returned speaker information.
• A newly available model is missing: Select Refresh models. Only model IDs beginning with muse-voice-transcribe- or muse-spark- are shown.

Cloud processing and privacy

Audio sent to Muse Voice Transcribe and text sent to Muse Spark are processed by Meta’s cloud API. Review Meta’s current terms, privacy information, regional availability, and pricing before using the add-on with sensitive content.