MiniMax Voice Cloning

Upload Clone Audio

Used to upload the original audio file for instant voice cloning.

POST /v1/files/upload

See File Management > Upload File, use voice_clone for purpose. Audio formats mp3/m4a/wav, duration 10 seconds to 5 minutes, ≤20 MB.

Upload Prompt Audio

Used to upload a prompt audio file to enhance the voice similarity and stability of speech synthesis.

POST /v1/files/upload

See File Management > Upload File, use prompt_audio for purpose. Audio formats mp3/m4a/wav, duration <8 seconds, ≤20 MB.

Instant Voice Cloning

Use this interface for instant voice cloning. If a cloned voice is not formally called within 7 days, the system will automatically delete it.

POST /v1/voice_clone

Request Headers

Parameter	Type	Required	Default	Description
Content-Type	string	Yes	application/json	Data exchange format
Authorization	string	Yes	-	Authentication info, format is Bearer <API_KEY>

Request Body

Parameter	Type	Required	Description
file_id	int64	Yes	The source audio file ID, obtained via the file upload interface
voice_id	string	Yes	Custom clone voice ID. Length [8, 256], must start with a letter, only letters/numbers/-/_ allowed, cannot end with - or _, and must not duplicate an existing ID
clone_prompt	object	No	Prompt audio, used to improve similarity and stability
clone_prompt.prompt_audio	int64	No	The file ID of the prompt audio
clone_prompt.prompt_text	string	No	The exact text of the prompt audio, which must fully match the audio content and end with punctuation
text	string	No	Preview synthesis text, up to 1000 characters. When provided, a preview audio link is returned and billed at the T2A rate
model	string	No	Preview synthesis model, required when text is present. Options: speech-2.8-hd, speech-2.8-turbo
language_boost	string	No	Enhance recognition of a specific language/dialect, or auto for automatic detection. Supports 40+ languages including Chinese, English, Japanese, Korean, Cantonese, etc.
text_validation	string	No	The expected transcription text of the source audio, up to 200 characters. The request is rejected when ASR similarity falls below the accuracy threshold (error code 1043)
accuracy	double	No	ASR similarity threshold, default 0.7, range [0, 1]. Used together with text_validation
need_noise_reduction	boolean	No	Whether to apply noise reduction to the source audio, default false
need_volume_normalization	boolean	No	Whether to enable volume normalization, default false
aigc_watermark	boolean	No	Whether to append an AIGC watermark beat marker at the end of the preview audio, default false

Request Example

curl --request POST \
  --url https://neolink.com/api/minimax/v1/voice_clone \
  --header 'Authorization: Bearer <API_KEY>' \
  --header 'Content-Type: application/json' \
  --data '{
    "file_id": 123456789,
    "voice_id": "my-custom-voice-001",
    "text": "你好，这是我的克隆音色试听。",
    "model": "speech-2.8-hd",
    "need_noise_reduction": true,
    "need_volume_normalization": true,
    "language_boost": "Chinese"
  }'

Successful Response (HTTP 200)

{
  "input_sensitive": {
    "type": 0
  },
  "demo_audio": "https://demo-audio.example.com/audio.wav",
  "extra_info": {
    "audio_length": 3500,
    "audio_sample_rate": 24000,
    "audio_size": 84000,
    "bitrate": 192000,
    "word_count": 14,
    "usage_characters": 14
  },
  "base_resp": {
    "status_code": 0,
    "status_msg": "success"
  }
}

Field	Type	Description
input_sensitive	object	Input content moderation result
input_sensitive.type	int	Moderation type: 0=normal, 1=serious violation, 2=pornographic, 3=advertisement, 4=prohibited, 5=abusive, 6=violent/terrorist, 7=other
demo_audio	string	Preview audio link (only returned when text and model are provided in the request)
extra_info	object	Preview audio metadata and billing info (only returned when text is provided in the request)
extra_info.audio_length	int64	Audio duration (milliseconds)
extra_info.audio_sample_rate	int64	Audio sample rate
extra_info.audio_size	int64	Audio size (bytes)
extra_info.bitrate	int64	Bitrate
extra_info.word_count	int64	Number of synthesized characters (excluding punctuation)
extra_info.usage_characters	int64	Number of billable characters
base_resp.status_code	int	Status code; see the error codes for specific definitions
base_resp.status_msg	string	Status details