Date: May 26, 2025

This document provides a comprehensive explanation of the API request methods supported by the TTS (Text-to-Speech) server.

📥 Clone Sound

Endpoint: POST /upload_clone
Body Type: form-data

Example

curl --location 'http://127.0.0.1:8000/upload_clone' \
--form 'voice_id="xiage"' \
--form 'ref_text="贝加尔湖是世界上最古老、最深的淡水湖泊，位于俄罗斯西伯利亚地区，湖水极其清澈透明，是世界上最纯净的湖泊之一。"' \
--form 'ref_audio=@"c:\\Users\\Joe\\Documents\\work\\AI_Avatar\\天印大道804号.m4a"'

Response

{
  "status": "success",
  "message": "Voice 'xiage' cloned successfully",
  "ref_text": "贝加尔湖是世界上最古老、最深的淡水湖泊，位于俄罗斯西伯利亚地区，湖水极其清澈透明，是世界上最纯净的湖泊之一。"
}

Note: ref_text must match the content of the audio in ref_audio.

📃 Query Voice List

Endpoint: POST /voices

Example

curl --location --request POST 'http://127.0.0.1:8000/voices'

Response

{
  "voices": [
    "xiaozhu",
    "xiage"
  ]
}

❌ Delete Voice

Endpoint: DELETE /voices/{voice_id}

Example

curl --location --request DELETE 'http://127.0.0.1:8000/voices/xiage'

Response

{
  "status": "success",
  "message": "Voice 'xiage' deleted successfully"
}

🗣️ Speech Synthesis

Endpoint: POST /tts
Body Type: application/json

Example

curl --location 'http://127.0.0.1:8000/tts' \
--header 'Content-Type: application/json' \
--data '{
  "text": "这是一条测试消息",
  "voice_id": "xiaozhu",
  "output_dir": "../outputs",
  "speed": 0.9,
  "nfe_step": 16,
  "cross_fade_duration": 0.15,
  "split_sentences": true,
  "remove_silence": true,
  "audio_format": "wav",
  "type": 1
}'

Response

{
  "job_id": "ec3e746b-3645-4c32-95df-225721d67d17",
  "status": "pending",
  "message": "TTS batch generation (from single tts with split_sentences) queued"
}

Notes:

  • output_dir: Custom output directory.

  • type: Indicates message type (e.g., chat, live comment). Default is none.

🛑 Cancel All TTS Tasks

Endpoint: POST /cancel_all_tts

Example

curl --location --request POST 'http://127.0.0.1:8000/cancel_all_tts'

Response

{
  "status": "success",
  "message": "Requested cancellation for 0 jobs",
  "cancelled_count": 0
}

Note: Used to clean up all running TTS tasks.

🔍 Query Job Status

Endpoint: GET /job_status/{job_id}

Example

curl --location --globoff 'http://127.0.0.1:8000/job_status/{job_id}'

Response

{
  "status": "completed",
  "submit_time": 1748272358.188642,
  "start_time": 1748272358.1892061,
  "end_time": 1748272368.3770256,
  "result": {
    "output_files": [
      "../outputs/priority_xiaozhu_20250526_151238_f6319314_part001.wav"
    ],
    "total_duration": 10.187206029891968,
    "voice_id": "xiaozhu",
    "sample_rate": 16000,
    "segments": 1,
    "type": 1,
    "job_id": "ec3e746b-3645-4c32-95df-225721d67d17"
  },
  "processing_duration": 10.187819480895996,
  "total_system_duration": 10.18838357925415
}

Note: Track progress and results using the job_id.

🚀 Start Smart Copy

Endpoint: POST /start_smartcopy
Body Type: application/json

Example

curl --location 'http://127.0.0.1:8000/start_smartcopy' \
--header 'Content-Type: application/json' \
--data '{
  "source_dir": "../outputs",
  "target_dir": "../stage"
}'

Response

{
  "status": "success",
  "message": "SmartCopy process started",
  "config": {
    "source_dir": "../outputs",
    "target_dir": "../stage"
  }
}

Note: Copies audio files from source to target using defined logic.

🛑 Stop Smart Copy

Endpoint: POST /stop_smartcopy

Example

curl --location --request POST 'http://127.0.0.1:8000/stop_smartcopy'

Response

{
  "status": "success",
  "message": "SmartCopy process stopped",
  "details": {
    "stdout": null,
    "stderr": null,
    "exit_code": -15
  }
}

🧹 Clear Directory

Endpoint: POST /clear_directory
Body Type: application/json

Example

curl --location 'http://127.0.0.1:8000/clear_directory' \
--header 'Content-Type: application/json' \
--data '{
  "directory": "../outputs"
}'

Response

{
  "status": "success",
  "message": "Cleared 60 files from '../outputs'",
  "files_cleared": 60,
  "errors": 0
}

Note: Deletes all files in the specified directory.