Everything on Inflect v3
1 mention, newest first.
Reddit user releases Inflect v3, a family of local text-to-speech models
A Reddit user reports releasing Inflect v3, a family of local text-to-speech models: Mini, Micro and Atom, under Apache 2.0. Mini is the main model, with six voices and 34.31M parameters. It provides vocal-event tags, reference voice-tuning that builds a profile from a recording, and 48 kHz enhancement.
Usable today