Latest / Kana & Mari's SoundRepos (English)

owenawsong/Inflect | local English text to speech engine | GitHub
Inflect v2 is an open-weight, local English text-to-speech system that generates complete 24 kHz waveforms without an external vocoder or hosted inference service. It provides compact Nano and Micro models with Python and CLI interfaces, CPU/CUDA inference, deterministic controls, punctuation-aware long-text synthesis, evaluation tooling, and optional comparison servers for ZipVoice and MOSS-TTS. Related: owenawsong/Inflect / local English text to speech engine / GitHub
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.