TTS

A FastAPI-based service for local, high-speed neural text-to-speech synthesis. Supports Piper and Kokoro backends.

demo.mp4

Installation

# Install system requirements
sudo apt install ffmpeg

# Install python dependencies
python3 setup.py
source tts-venv/bin/activate

Usage

Start the service:

python src/app.py

Python Example:

import requests

# Generate speech
response = requests.post(
    "http://localhost:47100/api/tts",
    json={
        "text": "Hello! {airhorn} Welcome.",
        "voice": "en_US-ryan-high"
    }
)

# Save the audio output
with open("speech.mp3", "wb") as f:
    f.write(response.content)

Endpoints

All routes are served under the /api prefix.

Synthesis

Method	Path	Description
POST	/api/tts	Synthesize speech from text, or from parts to concatenate text and SFX
GET	/api/sounds	List sound effects and aliases
GET	/sounds/{file}	Serve a sound file, mounted outside the /api prefix

Ops

Method	Path	Description
GET	/api/health	Backends, loaded voices, ffmpeg and cache stats

Catalog

Method	Path	Description
GET	/api/catalog	Voices, aliases, profiles and sounds in one payload
POST	/api/catalog/reload	Rescan voices and sounds and reread the blocklist
PUT	/api/catalog/{kind}/{name}	Create or update a voice_alias, profile or sfx_alias
DELETE	/api/catalog/{kind}/{name}	Delete a voice_alias, profile or sfx_alias

Moderation

Method	Path	Description
GET	/api/mod	Moderation terms, mode and censoring
POST	/api/mod	Set mode or censoring, add or remove terms, reload