Underdog Woof: a 31.7 MB TTS model that runs on CPU 13.8x faster than real time
victormustar · x · 2026-11-07
Underdog Woof is a tiny English TTS model (40.6 MB with two voices) that is 1022x smaller than Fish Audio Drama 3 (41.5 GB) while cutting word errors by 25%.
- Accuracy: 1.42% word error rate on Gradium TTS Hard Cases (English, 500 sentences), vs 1.89% for Fish Audio Drama 3
- Speed: 13.8x real time on a single Apple M5 Max CPU thread, using nothing but numpy — no GPU, server, or API key; peak memory 402 MB
- Details: built-in normalizer reads phone numbers, dates, prices, and codes the way people say them
- Ships with Juno and Mira voices; a 31.7 MB single-voice version runs in the browser. Licensed for non-commercial use.
More from Infra
- Satnam Singh: AI Is Dissolving the Hardware/Software Boundary for Good — satnam6502 · 2026-11-07
- Data centers' water use is just ~4% of US golf course irrigation, study argues — kimmonismus · 2026-11-07
- Data centers' water use vastly overstated: just 4% of US golf course irrigation — kimmonismus · 2026-11-07
- ByteDance reportedly weighing major expansion of AI data centers in Brazil — Polymarket · 2026-11-07
- Download an open-source model this weekend: one 5-20GB file, runs offline in ~20 min — alex_verem · 2026-11-07
- LLM inference flips between compute-bound and memory-bound within a single request — techNmak · 2026-11-07