Inflect-Micro-v2: complete voice in 9.36M parameters

Re:Inflect-Micro-v2: complete voice in 9.36M parameters by billdueber on Sunday July 26, 2026 @10:06AM (#49058380)

I keep seeing tts stories here. Is it just an interesting subset of the llm world, or is there a huge use case I’m somehow missing?

[ Parent | on HN ]

Re:Inflect-Micro-v2: complete voice in 9.36M parameters by eightysixfour on Sunday July 26, 2026 @11:34AM (#49059190)

I use STT/TTS to interface with a local LLM for Home Assistant in my house.

[ Parent | on HN ]

Re:Inflect-Micro-v2: complete voice in 9.36M parameters by _davide_ on Sunday July 26, 2026 @02:41PM (#49061021)

I'm do so as well, i tried qwen3 omni 3 but it was ridiculously stupid, and i ended up with stt thinker and tts. kokoro for now

[ Parent | on HN ]

Re:Inflect-Micro-v2: complete voice in 9.36M parameters by SamPatt on Sunday July 26, 2026 @09:59PM (#49064389)

I just built my own voice assistant with my Pebble Time 2 watch and it uses a VPS hosted TTS (Piper) and Hermes agent.

I learned about all of these projects on HN at one point or another.

[ Parent | on HN ]