click any column header to sort

TTS Bench — Speed — 2026-05-30_2241

Rig: windows-5090 — AMD Ryzen 9 9950X3D 16-Core Processor (16C) · NVIDIA GeForce RTX 5090 32GB · 126 GB RAM · Windows 11

Label: leaderboard-parity new models (default voice) — for merge into windows-default

TTFA = time to first audio (ms; lower is better). RTFx = realtime speed (× realtime; higher is better; e.g. 10× means 10 sec of audio generated per 1 sec of compute). Cold = first run after process start; warm = subsequent runs.

Speed winners

Fastest predefined-voice: Maya1 (cuda) — 0.59× warm RTFx, 14.80s warm TTFA

Fastest cloning-capable: StyleTTS 2 (cpu) — 33.76× warm RTFx, 251ms warm TTFA

Model	Device	TTFA cold	TTFA warm	RTFx cold	RTFx warm	Peak RAM	Peak VRAM	Size	Released
StyleTTS 2	cpu	1.74s	251ms	4.93×	33.76×	2.61 GB	—	~148M	Jun 2023
StyleTTS 2	cuda	1.76s	265ms	4.80×	32.15×	2.60 GB	1.49 GB	~148M	Jun 2023
OpenVoice v2	cuda	1.55s	347ms	4.05×	17.56×	—	1.35 GB	~100M	Apr 2024
OpenVoice v2	cpu	3.42s	1.31s	2.15×	4.73×	—	—	~100M	Apr 2024
VibeVoice 7B	cuda	5.67s	6.17s	1.39×	1.55×	2.05 GB	17.63 GB	7B	Sep 2025
Fish Speech 1.5	cuda	8.53s	7.84s	0.81×	0.94×	3.83 GB	1.80 GB	~500M	Nov 2024
Zonos v0.1	cuda	10.88s	10.38s	0.66×	0.71×	6.51 GB	4.48 GB	1.6B	Feb 2025
Maya1	cuda	16.57s	14.80s	0.53×	0.59×	2.42 GB	6.72 GB	3B	Oct 2025
Fish Speech 1.5	cpu	45.73s	45.34s	0.17×	0.17×	4.46 GB	—	~500M	Nov 2024
Zonos v0.1	cpu	62.12s	60.57s	0.12×	0.12×	7.39 GB	—	1.6B	Feb 2025
Maya1 (2/4 ok)	cpu	66.31s	73.66s	0.07×	0.07×	7.33 GB	—	3B	Oct 2025