xAI Grok Speech to Text

xAI Grok Speech to Text is xAI’s file-based transcription API. In Ayumi it writes a titled transcript without a prompt, with optional speaker labels, paragraph timestamps, filler words, text formatting and key-term biasing. You bring your own xAI API key.

Tip

Ayumi transcribes on-device without setup. Use xAI Grok when you want cloud transcription with diarization or vocabulary hints, then compare it with Gemini Transcribe, Muse Voice Transcribe, and Groq Whisper on your own recordings.

Getting an xAI API Key

1. Sign in to the xAI Console
2. Create an API key and make sure the account can use the API
3. Copy the key

Caution

Keep your API key private. Never share it publicly.

Configuring in Ayumi

1. Open Transcription Settings
2. Paste the key under xAI Grok
3. Star the built-in xAI Grok Speech to Text profile to make it the default, or pick it per recording

Every xAI Grok profile uses the base key under xAI Grok unless that profile sets an Override API Key. Keys are stored securely in your device’s Keychain and are not written into transcription tasks.

Options

Option	What it does
Language	Leave it on Automatic for detection. Choosing one of the 25 listed languages lets xAI format spoken numbers, currencies and units.
Format Numbers & Units	Applies xAI’s inverse text normalization. This option is available only when a language is selected.
Label Speakers	Writes one paragraph per detected speaker turn (**Speaker 1:** …).
Word Timestamps	Adds a [mm:ss–mm:ss] range to each speaker turn, or to sentence-sized paragraphs when speaker labels are off.
Include Filler Words	Keeps words such as “um” and “uh”; xAI removes them by default.
Key Terms	Biases names, jargon and product names from a newline-separated Markdown file, terms typed into the profile, or both.

Key terms are sent in file-first order, with duplicates removed. xAI accepts up to 100 terms of 50 characters each; Ayumi trims the list to those limits before upload. The same Markdown vocabulary file can be reused by the other cloud transcript engines.

Uploads and Limits

• Ayumi sends its AAC-in-M4A recording directly to xAI’s batch endpoint; there is no lossy intermediate conversion.
• xAI accepts files up to 500 MB. Ayumi rejects a larger file before upload.
• A network connection is required for each transcription.
• Ayumi currently uses file transcription, not xAI’s live WebSocket streaming mode.

See the xAI Speech to Text documentation for the current API formats, limits and behavior.

Data Privacy

When using an xAI Grok profile:

• Audio is sent directly from your device to xAI for processing
• Ayumi shows its cloud-processing consent dialog before the first cloud transcription
• No audio passes through or is stored on Ayumi’s servers
• See xAI’s terms and policies for xAI’s handling of API data