How hexgrad/kokoro works
Kokoro is an inference library for multilingual text-to-speech, with Python and JavaScript implementations. The main Python workflow accepts text through KPipeline, applies language-aware phonemization and voice selection, then invokes KModel to synthesize audio; users can also use the CLI to write WAV output. The model combines text/prosody components with an audio decoder and STFT support. The JavaScript API provides a parallel generate workflow with sentence splitting, phonemization, voice data, and a Transformers runtime. Relationships for unsampled internals are omitted; documented capabilities and explicit imports are retained.
Main entry points: pipeline.py KPipeline · kokoro.js KokoroTTS API · model.py KModel · istftnet.py Audio decoder
Components 17
Python API
- CLI WAV output
kokoro/__main__.py - KPipeline
kokoro/pipeline.py - Language G2P
kokoro/pipeline.py - Voice loading
kokoro/pipeline.py
Python Model
- KModel
kokoro/model.py - Text and prosody nets
kokoro/modules.py - Audio decoder
kokoro/istftnet.py - Custom STFT
kokoro/custom_stft.py
JavaScript Runtime
- KokoroTTS API
kokoro.js/src/kokoro.js - Text phonemizer
kokoro.js/src/phonemize.js - Sentence splitter
kokoro.js/src/splitter.js - Voice data
kokoro.js/src/voices.js
External Services
- Transformers runtime
- Misaki G2P
- Hugging Face Hub
Other components
- Text caller
- JS API caller
Connections 18
- Text caller → CLI WAV output: provides text
kokoro/__main__.py - CLI WAV output → KPipeline: generates audio
kokoro/__main__.py - KPipeline → Language G2P: phonemizes text
kokoro/pipeline.py - KPipeline → Voice loading: loads voice
kokoro/pipeline.py - KPipeline → KModel: invokes model
kokoro/pipeline.py - KPipeline → Misaki G2P: uses G2P
kokoro/pipeline.py - Voice loading → Hugging Face Hub: downloads voice
kokoro/pipeline.py - KModel → Hugging Face Hub: loads weights
kokoro/model.py - KModel → Text and prosody nets: uses networks
kokoro/model.py - KModel → Audio decoder: uses decoder
kokoro/model.py - Text and prosody nets → Audio decoder: uses blocks
kokoro/modules.py - Audio decoder → Custom STFT: uses transform
kokoro/istftnet.py - JS API caller → KokoroTTS API: generates speech
kokoro.js/src/kokoro.js - KokoroTTS API → Text phonemizer: phonemizes text
kokoro.js/src/kokoro.js - KokoroTTS API → Sentence splitter: splits text
kokoro.js/src/kokoro.js - KokoroTTS API → Voice data: loads voice data
kokoro.js/src/kokoro.js - KokoroTTS API → Transformers runtime: runs model
kokoro.js/src/kokoro.js - Text phonemizer → Transformers runtime: uses phonemizer
kokoro.js/src/phonemize.js