feat: add KittenTTS text-to-speech model#118

Closed

beshkenadze wants to merge 12 commits intoBlaizzy:mainfrom

beshkenadze:feat/kitten-tts

Commits on Mar 22, 2026

feat: add MLXAudioModules shared NN blocks and MLXAudioG2P grapheme-to-phoneme
beshkenadze
committed
refactor: move CMUdict to HF download, remove bundled resources
beshkenadze
committed

Commits on Mar 23, 2026

fix: address Copilot review — rename SmokeTests, reorder normalizer rules, remove stale doc link
beshkenadze
committed
chore: remove AGENTS.local.md from repo
beshkenadze
committed
chore: add AGENTS.local.md to .gitignore
beshkenadze
committed
fix: remove version constant from MLXAudioG2P namespace
beshkenadze
committed
refactor: move MLXAudioModules into Models/StyleTTS2/Blocks, remove as separate package
beshkenadze
committed
feat: add MLXAudioNeuralG2P ByT5 multilingual grapheme-to-phoneme
beshkenadze
committed
fix: address Copilot review — trim phonemize output, pattern-match ignored keys, use MLX range, update scheme list
beshkenadze
committed
feat: add shared TTS types, G2P stack, and text processing
beshkenadze
committed
fix: throwing init for EnglishFallbackNetwork + add prepare() to TextProcessor protocol
beshkenadze
committed
feat: add KittenTTS text-to-speech model with Module-based architecture
beshkenadze
committed