Quickstart

Install KittenTTS and generate your first audio in minutes.

KittenTTS runs on CPU only. No GPU is required or supported.

1. Install the package

   Install KittenTTS directly from the GitHub release:

   pip install https://github.com/KittenML/KittenTTS/releases/download/0.8.1/kittentts-0.8.1-py3-none-any.whl
   
2. Load a model

   KittenTTS offers three model sizes. Choose based on your quality and speed requirements:

   from kittentts import KittenTTS
   
   model = KittenTTS("KittenML/kitten-tts-mini-0.8")   # 80M — highest quality
   # model = KittenTTS("KittenML/kitten-tts-micro-0.8") # 40M — balanced speed and quality
   # model = KittenTTS("KittenML/kitten-tts-nano-0.8")  # 15M — smallest and fastest
   

   Models are downloaded from Hugging Face on first use and cached locally. You can specify a custom cache directory:

   model = KittenTTS("KittenML/kitten-tts-mini-0.8", cache_dir="/path/to/cache")
   
3. Generate audio and save to file

   Generate speech from text and save the result as a WAV file:

   import soundfile as sf
   from kittentts import KittenTTS
   
   model = KittenTTS("KittenML/kitten-tts-mini-0.8")
   
   text = """One day, a little girl named Lily found a needle in her room. She knew it was difficult to play with it because it was sharp."""
   
   audio = model.generate(text=text, voice="Bruno")
   sf.write("output.wav", audio, 24000)
   

   generate() returns a NumPy array at 24 kHz. You can use any audio library that accepts NumPy arrays, or use generate_to_file() to skip the intermediate step:

   model.generate_to_file(text, "output.wav", voice="Bruno")
   
4. List available voices

   Check which voices are available on your loaded model:

   print(model.available_voices)
   # ['Bella', 'Jasper', 'Luna', 'Bruno', 'Rosie', 'Hugo', 'Kiki', 'Leo']
   

Advanced usage

Adjust speed

# Speed > 1.0 is faster, < 1.0 is slower
audio = model.generate("Hello, world.", voice="Luna", speed=1.2)

If you omit the voice argument, KittenTTS uses Leo by default. Pass any name from model.available_voices to switch voices.

Next steps

How it works

Learn about the ONNX runtime and how KittenTTS processes text into audio.

Voices

Explore the available voices and how to choose the right one for your use case.

Models overview

Compare model sizes and understand the trade-offs between quality and speed.

API reference

Full reference for the KittenTTS class, parameters, and return values.