MiniMax Voice Design · API reference
Describe a virtual voice, play its preview, and reuse it for text-to-speech synthesis.
Send a JSON request to POST /v1/wand/minimax-tts/sync_design with an APIMaster API key in the Authorization: Bearer header. Voice Design creates a virtual voice from a text description.
| Field | Type | Description |
|---|---|---|
model |
string | Required. Use minimax-voice-design. |
prompt |
string | Required. Describe the voice, delivery, emotion, and intended use. |
preview_text |
string | Required. Up to 500 characters of text for the audio preview. |
voice_id |
string | Optional custom voice ID. An ID is generated if omitted. |
output_format |
string | hex (default) or url. Preview URLs last 24 hours. |
aigc_watermark |
boolean | Append an audio watermark. Defaults to false. |
curl -X POST "https://apimaster.ai/v1/wand/minimax-tts/sync_design" \
-H "Authorization: Bearer $APIMASTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"minimax-voice-design","prompt":"A warm and clear narrator at a natural pace","preview_text":"Welcome to the program.","output_format":"url"}'
Check base_resp.status_code=0 and use base_resp.status_msg to diagnose errors. Play the preview in trial_audio as hex-decoded audio or a URL. Keep request_id or trace_id if a request fails. The response can include usage.total_token.
Pass the returned voice_id unchanged as voice_setting.voice_id in MiniMax Speech using the same account. Preview and subsequent synthesis are metered separately. The first full synthesis may include a voice fee. An unused temporary voice expires after seven days. See the Voice Design model card for the current APIMaster price.
