Text-to-speech for AI agents.
54+ voices, 19 emotions, SSML, voice cloning, and an MCP server with 40+ tools. One pip install, runs locally on your machine.
Python
engine = VoiceEngine()
result = engine.speak('Hello!', emotion='excited')
CLI
voice-soundboard speak 'Hello world!'
# -> output/hello_world.wav
MCP
# Ask Claude: "Say hello in an excited voice"
# -> generates audio via 40+ MCP tools
Features
Everything you need for speech synthesis.
54+ voices
Male, female, and character voices with British, American, and international accents via Kokoro TTS.
19 emotions
Happy, sad, angry, excited, calm, fearful, nostalgic, determined — fine-grained emotional control.
MCP server
40+ tools for AI agents. Add to Claude Desktop and ask it to speak — it just works.
Quick start
Install & speak
pip install voice-soundboard
from voice_soundboard import VoiceEngine
engine = VoiceEngine()
result = engine.speak('Hello world!')
print(result.audio_path)MCP for Claude
{
"mcpServers": {
"voice-soundboard": {
"command": "python",
"args": ["-m", "voice_soundboard.server"]
}
}
}TTS Engines
Kokoro ships by default. Add more with optional installs.
Advanced
Beyond the basics.
Voice cloning
Clone any voice from a short audio sample with F5-TTS. Requires explicit consent acknowledgment.
SSML support
Fine-grained control with Speech Synthesis Markup Language — pauses, emphasis, prosody.
Multi-speaker
Generate conversations with multiple voices. Script dialogues with per-character emotion and style.