Skip to content
Ticos Docs

TTS Speaker Training

Basics

  • Base URL: https://api.ticos.cn
  • Train: POST /train_speaker
  • Check status: GET /train_speaker
  • Delete: DELETE /train_speaker

Supported models

  • tts-turbo
  • tts-turbo-v2
  • tts-pro

1. Train speaker (POST)

Request parameters (multipart/form-data)

  • audio (required): WAV file
  • model (required): TTS model
  • audio_text (required): transcript of the uploaded sample

Example

curl --location 'https://api.ticos.cn/train_speaker' \
--form 'audio=@"/path/to/audio.wav"' \
--form 'model="tts-turbo"' \
--form 'audio_text="Hello, this is a training sample."'

Success response

{
  "speaker_id": "tispk_s1_xxx",
  "code": 0,
  "message": "create success"
}

2. Check training status (GET)

Query parameters

  • model (required)
  • speaker_id (required)

Example

curl --location 'https://api.ticos.cn/train_speaker?model=tts-turbo&speaker_id=tispk_s1_xxx'

Response example

{
  "model": "tts-turbo",
  "speaker_id": "tispk_s1_xxx",
  "status": 1
}
  • status=1: completed
  • status=0: in progress
  • status=-1: speaker not found

3. Delete speaker (DELETE)

Query parameters

  • model (required)
  • speaker_id (required)

Example

curl --location --request DELETE 'https://api.ticos.cn/train_speaker?model=tts-turbo&speaker_id=tispk_s1_xxx'

Response example

{
  "code": 0,
  "message": "voice delete success"
}

Notes

  • Use clean WAV samples with minimal noise.
  • Keep audio_text aligned with the recording content.
  • Check status before destructive operations.