How kittenml/kitten-tts-2-cpp works
This fork’s README describes KittenTTS2 as a batch CPU speech-synthesis CLI: users provide text and voice/options, the runtime prepares or downloads mode-specific assets, generates tokens, decodes waveforms, and writes WAV output. The tree does not include the documented Kitten-specific runtime files, so those nodes use null paths; llama.cpp and GGML paths are included where the tree supports them. Edges are based on README statements, not an inspection of unsampled runtime code.
Main entry points: llama.h llama.cpp runtime · llama-context.cpp Inference context · ggml.h GGML compute
Components 19
Speech workflow
- TTS CLI
- Text normalization
- Sentence chunking
- Sampling controls
- Text-processing submodule
Model assets
- Model repository
- Asset downloads
- Asset cache
- GGUF model
- Voice conditioning
- Decoder exports
Inference runtime
- llama.cpp runtime
include/llama.h - Inference context
src/llama-context.cpp - GGML compute
ggml/include/ggml.h
Audio output
- S3 waveform decoder
- Audio joins
- 24 kHz WAV
Other components
- CLI user
- CPU LibTorch
Connections 18
- CLI user → TTS CLI: provides options
README.md - TTS CLI → Asset downloads: prepares assets
README.md - Asset downloads → Model repository: fetches assets
README.md - Asset downloads → Asset cache: caches assets
README.md - TTS CLI → Text normalization: normalizes text
README.md - Text normalization → Text-processing submodule: uses
README.md - TTS CLI → Sentence chunking: chunks sentences
README.md - TTS CLI → Sampling controls: configures sampling
README.md - TTS CLI → llama.cpp runtime: runs inference
README.md - Asset cache → GGUF model: supplies model
README.md - Asset cache → Voice conditioning: supplies conditioning
README.md - Asset cache → Decoder exports: supplies decoder
README.md - llama.cpp runtime → Inference context: provides context API
src/llama-context.cpp - Inference context → GGML compute: computes graphs
src/llama-context.cpp - TTS CLI → S3 waveform decoder: decodes waveforms
README.md - S3 waveform decoder → CPU LibTorch: uses
README.md - S3 waveform decoder → Audio joins: produces chunks
README.md - Audio joins → 24 kHz WAV: writes audio
README.md