click any column header to sort

Published Runs

Open-source TTS models benchmarked side-by-side: speed (TTFA, RTF) and voice cloning. Each run has inline audio so you can listen to every model × prompt without downloading. Repo on GitHub →
6 published run(s)
Run Label Rig Models Devices Prompts Rows OK Report
windows-default default voice windows-5090 27 models cpu, cuda 5 607 600/607 speed · samples
windows-cloning cloning · chris_hemsworth_15s windows-5090 19 models cpu, cuda 5 423 411/423 speed · samples
mac-default default voice mac-m4 23 models cpu, mps 5 418 387/418 speed · samples
mac-cloning cloning · jo mac-m4 15 models cpu, mps 5 264 236/264 speed · samples
linux-default default voice linux-3090 26 models cpu, cuda 5 606 600/606 speed · samples
linux-cloning cloning · chris_hemsworth_15s linux-3090 17 models cpu, cuda 5 402 396/402 speed · samples