owenawsong/Inflect: Inflect v3 local TTS model

Inflect v3 is a new family of compact local text-to-speech models ranging from 157K to 34M parameters, offering native 24 kHz speech generation, reference-voice tuning, and optional vocal event tags. Available under an Apache 2.0 license, the models can run locally on CPU or optimized CUDA hardware without a hosted API. Published benchmarks report high intelligibility and real-time generation throughput on supported hardware. Open article →