Model API

MiMo 2.5 TTS

Expressive speech from text with Xiaomi MiMo 2.5. Choose a preset voice, describe a new one, or clone your own voice from a recording, and direct the delivery with tags like (whispering) and [pause]. For narration, audiobooks, game characters, assistant replies and video voiceovers.

Model id

Use this in every request

mimo-2-5-tts

Run it

Quote the input, start the generation with the quote, then poll the job. Set RAYWAKE_API_KEY first; see Authentication.

# 1. Price the exact input (quotes are valid for up to 10 minutes)
curl -s https://api.raywake.com/v1/quotes \
  -H "Authorization: Bearer $RAYWAKE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "mimo-2-5-tts",
  "input": {
    "prompt": "A red ceramic cup on a wooden table, soft morning light"
  }
}'

# 2. Start the generation with the quote_id from step 1
curl -s https://api.raywake.com/v1/generate \
  -H "Authorization: Bearer $RAYWAKE_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $(uuidgen)" \
  -d '{"model": "mimo-2-5-tts", "quote_id": "QUOTE_ID", "input": {"prompt":"A red ceramic cup on a wooden table, soft morning light"}}'

# 3. Poll until the status is terminal
curl -s https://api.raywake.com/v1/jobs/JOB_ID \
  -H "Authorization: Bearer $RAYWAKE_API_KEY"

Input

Input fields for mimo-2-5-tts
Name	Type	Description
`prompt`required	string	The text to speak, up to about 8,000 tokens per request; split longer scripts. Start with a style tag in parentheses, such as (whispering) or (singing), and add inline tags in brackets where they should happen, such as [pause], [laughing] or [sigh]. A misformatted tag may be read aloud. Singing works only with a built-in voice. Example: Close your eyes, take a deep breath, [pause] and let the day slowly drift away.
`voice`	string (enum)	Optional. Default Mia. The built-in voice that speaks the text. Ignored when a Voice Sample or Voice Description is given. Default: `Mia` One of: `Mia`, `Chloe`, `Milo`, `Dean`, `冰糖`, `茉莉`, `苏打`, `白桦`
`voiceDescription`	string	Optional. Creates a new voice from a 1-4 sentence description: gender and age, texture, mood and pace, e.g. "A deep, gravelly old man voice, slow and warm." Used instead of the built-in voice. Ignored when a Voice Sample is given.
`inputAudio`	array of image URLs	Optional. Clone a voice from a recording; the text is spoken in that voice. Use only your own voice or one you have permission to use. A clear, single-speaker recording without music, noise or echo works best; MP3 or WAV. Takes priority over Voice and Voice Description. Singing is not available with a cloned voice. Default: `[]`
`instructions`	string	Optional. Directions for how the text is spoken, such as tone, emotion, pace or character, e.g. "Warm, calm narrator tone." Not spoken.
`outputFormat`	string (enum)	Optional. Default mp3. Audio format of the output file. Default: `mp3` One of: `mp3`, `wav`

Output

A succeeded job lists its results under outputs, stored on our side. See Jobs for every field.

Job excerpt

"outputs": [
  { "type": "audio", "url": "https://…", "mime_type": "audio/mpeg", "thumbnail_url": null }
]

Pricing

FreeFree/ request

The exact charge for your input comes from a quote. How charges settle