Speech - MountSea API

Speech is not a new task. It is task=create with is_speech: true, and it is billed as create. Omit is_speech (or send false) and the call is a normal song — the field is not sent upstream. is_speech and backing_music only apply when true, and only on create. Official captures use chirp-v6-mini. POST /suno/v2/generate returns { "taskId" }. Poll Get Task Status.

Parameters

Field	Required	Notes
`task`	Yes	`"create"`
`model`	Yes	`chirp-v6-mini` in the official capture. `chirp-v6` is also accepted
`is_speech`	Yes for speech	`true`
`prompt`	Script form	The words to speak. Non-empty → Custom
`tags`	Script form	Voice style: pace, tone, “no whisper”
`gpt_description_prompt`	Idea form	The idea. Omit `prompt` and `persona`
`vocal_gender`	No	`"f"` or `"m"`. Omit = no preference
`backing_music`	No	`true` = a music bed under the voice. Leave it unset on the idea form unless you want that bed
`controls.aug_creativity`	No	Variety, 0–4. `4` = max

Examples

Write the script. prompt is the words. tags is how it should sound:

{
  "task": "create",
  "model": "chirp-v6-mini",
  "prompt": "一张照片，一首歌，一支专属 MV。\n上传你的形象，选好音乐，几分钟就能生成电影感音乐视频。",
  "tags": "温暖女声，吐字清晰，语速中等偏慢，亲和有自信，不要耳语也不要喊叫",
  "vocal_gender": "f",
  "is_speech": true,
  "backing_music": true,
  "controls": { "aug_creativity": 4 }
}

Idea only. Do not send prompt, persona, or backing_music unless you want a bed:

{
  "task": "create",
  "model": "chirp-v6-mini",
  "gpt_description_prompt": "A fiery halftime locker-room speech from a coach whose team is down twenty points, building from a low growl to a full-throated roar",
  "vocal_gender": "f",
  "is_speech": true
}

​Parameters

​Examples

Parameters

Examples