Pricing is temporarily unavailable — capabilities and docs are still fully viewable.
Billing details
Description
Xiaomi MiMo v2.5 TTS VoiceClone — voice cloning model: supply a reference audio sample as a data URL in the voice field (data:{mime};base64,...) and it synthesises speech in that voice with no training, annotation, or fine-tuning step. Accepts MP3 (audio/mpeg) or WAV, with the base64-encoded string capped at 10 MB; Xiaomi publishes no minimum reference duration. Natural-language style instructions and inline audio tags carry over from mimo-v2.5-tts, but streaming is not available — the request runs in compatibility mode and returns once synthesis has completed.
Capabilities
- Tasks
- text-to-speech, voice-cloning
API endpoints
POST /v1/audio/speech(audio-speech)