KittenML/KittenTTS

Features

• Neural Text-to-Speech Engines - Uses lightweight neural network models to map text directly to audio waveforms for natural speech synthesis.
• Text-to-Audio Synthesis - Converts written text into audio samples using lightweight models with adjustable playback speed.
• Text-to-Speech - Synthesizes spoken audio from written text using a neural network model optimized for speed.
• Speech Synthesis Formatters - Prepares raw text by expanding abbreviations and numbers to ensure high-quality synthesized speech.
• Text-to-Speech Normalizers - Processes raw text to expand numbers and abbreviations into full spoken words for more natural synthesis.
• Automated Content Creation Tools - Generates audio files from text scripts to create voiceovers without manual recording.
• Lightweight Voice Generation - Produces spoken audio with adjustable speed using models optimized for low-resource hardware.
• Model Quantization - Uses model quantization to reduce precision, lowering memory usage and increasing inference speed on consumer hardware.
• Prosody Controls - Allows adjustment of speech speed and tone by modifying input variables during the inference process.
• Audio Generation - Functions as a generator that writes synthesized speech directly to audio files.
• Audio Exporters - Provides the ability to serialize synthesized audio buffers into standard files for permanent storage.
• Generative Audio Chunking - Implements sequential yielding of audio chunks during synthesis to enable low-latency playback.
• Audio and Video Processing - Provides lightweight, CPU-friendly text-to-speech synthesis.
• Audio Video Processing - Listed in the “Audio Video Processing” section of the Awesome Python awesome list.
• Speech Processing - Lightweight text-to-speech synthesis.

Star history