Phonon TTS and Fluid Speech Mini: a day-long test of lightweight local speech models
gregbarbosa · x · 2026-10-22
User gregbarosa reports using Phonon TTS and Fluid Speed Mini STT all day at work and calling them his new favorite local speech models — fast, lightweight, and does exactly what he needs.
Screenshots show both engines integrated as Beta options in a Mac dictation/speech tool: the Phonon TTS model downloads once and runs locally, while Fluid Speech Mini (CC BY 4.0, hosted on Hugging Face) is a fast, small STT model that ignores other voices, paired with cleanup/rewrite post-processing options.
More from Infra
- BigQuery ObjectRef goes GA: query unstructured data in SQL with Gemini — rseroter · 2026-10-22
- Why we invested in Daytona: autonomous agents will devour sandbox compute — pzakin · 2026-10-22
- Image-to-video in ComfyUI without a local GPU: Forma custom nodes, pay-as-you-go from $5 — no3us · 2026-10-22
- Meta's Muse Spark Open: 816B-parameter open-weight model, 1M context, live on Together — togethercompute · 2026-10-22
- Speculative data center debt repackaged and sold to pension funds, echoing subprime bubble — churchkey · 2026-10-22
- Running Qwen3.8 Flash Next on 4x RTX 3090s: Which Inference Engine? — Civil_Fee_7862 · 2026-10-22