task_id, then poll for the result.custom_mode:custom_mode: true — your script. prompt is read aloud as written. Use tags to direct the voice: tone, pacing, emotion, scene (for example calm, warm, steady pace, evening news).custom_mode: false — from a description. A script is written and read from gpt_description_prompt (for example Read an evening news segment in a calm male voice at a steady pace). tags, background_music, vocal_gender and variety are not accepted in this mode.task_id in the responseGET /api/v1/sonic/task/{task_id} until the task completes, or pass webhook_urlaudio_url, duration, title and tags{
"code": 200,
"data": [
{
"clip_id": "xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx",
"state": "succeeded",
"title": "Good Evening",
"tags": "Warm mid-range female voice, calm evening newsreader, steady measured pace",
"lyrics": "Good evening. Here is tonight's news.",
"audio_url": "https://files.musicapi.ai/media/xxxxxxxx-audio_url.mp3",
"duration": 13.6
},
{
"clip_id": "yyyyyyyy-yyyy-yyyy-yyyy-yyyyyyyyyyyy",
"state": "succeeded",
"title": "Good Evening",
"audio_url": "https://files.musicapi.ai/media/yyyyyyyy-audio_url.mp3",
"duration": 13.2
}
],
"message": "success"
}