Custom Voices

Create Professional Clone

SDK

Use the SDK for safer, faster integration than manual HTTP calls.

Starts asynchronous professional voice training from WAV or MP3 recordings. Send the recordings, display name, supported TTS model, and an ISO 639-3 language code as multipart/form-data.

Audio requirements

One WAV or MP3 file
File size: 1 GiB or less
Duration: 5 minutes to 3 hours
Sample rate: 16 kHz or higher

A successful request returns 202 Accepted with status: training. Poll GET /v1/custom-voices/{voice_id} until the status becomes completed or failed.

Training may take up to two hours. You’ll receive an email when custom voice training is complete or if it fails.

POST/v1/custom-voices/professional-clone

cURL

curl --request POST \
  --url https://api.typecast.ai/v1/custom-voices/professional-clone \
  --header 'X-API-KEY: <api-key>' \
  --form 'files=@training-audio.wav' \
  --form 'name=Custom Voice Name' \
  --form 'model=ssfm-v30' \
  --form 'language=eng'

{
  "name": "Custom Voice Name",
  "model": "ssfm-v30",
  "status": "training",
  "voice_id": "uc_6700000000000000000000bb"
}

Authorizations

X-API-KEYstringheaderrequired

API key for authentication. You can obtain an API key from the Typecast API Console.

Body

multipart/form-data

Multipart request body for professional voice cloning.

namestringrequired

Voice name, up to 30 characters.

filesstring[]required

WAV or MP3 recordings used for training. Only one file can be uploaded.

Upload limits

One WAV or MP3 file
File size: 1 GiB or less
Duration: 5 minutes to 3 hours
Sample rate: 16 kHz or higher

For best results, we recommend audio that meets the following conditions:

Record in a speaking style that closely matches how you want the generated voice to sound.
Record in a quiet environment without background noise.
Include only one speaker.
Record in the language specified in the language field.
Longer input audio results in higher-quality generated voices.

modelssfm-v30 | ssfm-v21required

TTS model version.

ssfm-v30ssfm-v21

languagestringrequired

ISO 639-3 language code, such as kor or eng.

Response

202Professional clone training started successfullyapplication/json

namestring

Voice name.

modelstring

TTS model version.

statuspending | training | completed | failed

Current creation or training status.

pendingtrainingcompletedfailed

voice_idstring

Unique custom voice identifier with the uc_ prefix.

cURL

curl --request POST \
  --url https://api.typecast.ai/v1/custom-voices/professional-clone \
  --header 'X-API-KEY: <api-key>' \
  --form 'files=@training-audio.wav' \
  --form 'name=Custom Voice Name' \
  --form 'model=ssfm-v30' \
  --form 'language=eng'

{
  "name": "Custom Voice Name",
  "model": "ssfm-v30",
  "status": "training",
  "voice_id": "uc_6700000000000000000000bb"
}