Skip to main content
POST

Voice presets (input.voice)

Pass the voice ID in input.voice (maps to MiniMax voice_id). The table lists 327 system voices.

Language (input.language)

Optional. Maps to MiniMax language_boost to improve synthesis for a target language or dialect. Omit or set auto for automatic detection.
For Cantonese text, set language to Chinese,Yue and pick a Cantonese voice preset (e.g. Cantonese_*).
Custom voice_id strings (up to 64 characters) are supported when voice cloning is enabled on your account.

Optional parameters

Authorizations

Authorization
string
header
required

Authorization: Bearer YOUR_API_KEY

Body

application/json
model
string
default:speech-2.8-hd
required

Model ID. Use speech-2.8-hd for this endpoint.

Example:

"speech-2.8-hd"

input
object
required
webhook
string<uri>

Optional HTTPS callback URL when the task completes, fails, or is cancelled.

Response

Async task created

data
object
required