Text-to-Speech: Convert text to speech using Edge-TTS (Russian female voice)
Voice Conversion: Apply voice cloning using Seed-VC model
Reference Voices: Upload and manage reference voice samples
OpenAPI/Swagger: Interactive API documentation
Token-based Auth: Secure API access

API Endpoints

Endpoint	Method	Description	Auth
`/health`	GET	Health check	No
`/voices`	GET	List all voices	Yes
`/voices/upload`	POST	Upload new voice	Yes
`/tts`	POST	Convert text to speech	Yes

Quick Start

1. Clone and Setup

git clone https://git.azimkin.dev/Azimkin/tts-vc-api.git
cd tts-vc-api
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

2. Configure

Copy .env.example to .env and configure:

cp .env.example .env

Edit .env:

API_TOKENS=your-secret-token
VOICES_DIR=./voices
OUTPUT_DIR=./output
EDGE_TTS_VOICE=ru-RU-SvetlanaNeural
FLASK_PORT=5000

3. Run

python run.py

API will be available at http://localhost:5000 Swagger UI: http://localhost:5000/swagger-ui

Usage Examples

Health Check

curl http://localhost:5000/health

Upload Reference Voice

curl -X POST http://localhost:5000/voices/upload \
  -H "Authorization: Bearer your-token" \
  -F "file=@reference.wav" \
  -F "name=My Voice"

List Voices

curl http://localhost:5000/voices \
  -H "Authorization: Bearer your-token"

Text-to-Speech

curl -X POST http://localhost:5000/tts \
  -H "Authorization: Bearer your-token" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Привет, мир!",
    "voice_id": "voice-id-from-upload",
    "speed": 1.0
  }' \
  --output output.wav

Architecture

project/
├── app/
│   ├── api/           # API endpoints (Flask-RESTX)
│   ├── services/      # Business logic (TTS, VC)
│   ├── models/        # Data models
│   └── config.py      # Configuration
├── voices/            # Reference voices storage
├── output/            # Generated audio temp storage
└── run.py             # Entry point

Tech Stack

Framework: Flask + Flask-RESTX
TTS: edge-tts
Voice Conversion: Seed-VC (seed-uvit-whisper-small-wavenet)
Auth: Token-based from .env

Environment Variables

Variable	Description	Default
`API_TOKENS`	Comma-separated valid tokens	-
`VOICES_DIR`	Reference voices directory	`./voices`
`OUTPUT_DIR`	Output audio directory	`./output`
`EDGE_TTS_VOICE`	Edge-TTS voice	`ru-RU-SvetlanaNeural`
`FLASK_PORT`	Server port	`5000`
`FLASK_DEBUG`	Debug mode	`false`

License

MIT