About Kitten TTS: Open-source (Apache 2.0) by KittenML. ONNX-based, 15M-80M params, runs on CPU. v0.8.1.

Prerequisites

  • Windows 10/11 64-bit
  • Python 3.8+
  • pip
  • NVIDIA GPU optional (CPU works great)
  • ~100 MB disk space

Step 1: Install Python

Download Python 3.10+ from python.org. Check "Add Python to PATH".

python --version

Step 2: Virtual Environment

mkdir kitten-tts-project
cd kitten-tts-project
python -m venv venv
venv\Scripts\activate

Step 3: Install Kitten TTS v0.8.1

pip install soundfile
pip install https://github.com/KittenML/KittenTTS/releases/download/0.8.1/kittentts-0.8.1-py3-none-any.whl

Step 4: Verify

python -c "from kittentts import KittenTTS; print('OK!')"

Step 5: Generate First Audio

from kittentts import KittenTTS
import soundfile as sf

model = KittenTTS("KittenML/kitten-tts-mini-0.8")
audio = model.generate("Hello from Kitten TTS on Windows!", voice="Jasper")
sf.write("output.wav", audio, 24000)
print("Saved output.wav")

Step 6: Try All 8 Voices

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

for voice in model.available_voices:
    model.generate_to_file(f"Hi, I am {voice}.", f"{voice}.wav", voice=voice)

Optional: GPU Acceleration

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
model = KittenTTS("KittenML/kitten-tts-mini-0.8", backend="cuda")

Model Variants

ModelParamsSizeBest For
kitten-tts-mini-0.880M80 MBBest quality production
kitten-tts-micro-0.840M41 MBBalanced speed/quality
kitten-tts-nano-0.815M56 MBFastest, edge devices
kitten-tts-nano-0.8-int815M25 MBUltra-lightweight

Troubleshooting

CUDA not detected: Update drivers. Check: python -c "import torch; print(torch.cuda.is_available())"
Memory error: Try nano (15M) or micro (40M) models instead.