benchmarks.site

Public leaderboard · Text-to-speech

TTS Arena V2

People listen to two computer-generated voices reading the same text and pick the one that sounds better. The ranking shows which text-to-speech systems people actually prefer.

Arena (people vote)

Top 3 right now

  1. 1stAurora1687
  2. 2ndstar-june-20261633
  3. 3rdInworld TTS MAX1567

What it measures

Blind pairwise listening votes between text-to-speech models on the same text, converted to Elo ratings.

How to read the numbers

Elo rating from blind human votes (higher = preferred more often) — higher is better.

Results

25 rows · higher is better (Elo)CSV
1Aurora168773.4312
2star-june-2026163364.7295
3Inworld TTS MAX156763.81193
4CastleFlow v1.0155861.91490
5Inworld TTS 1.5 MAX156559.7390
6Hume Octave155159.7949
7Inworld TTS154360.01712
8Papla P1155261.2836
9Gradium TTS 2026.10158862.7102
10Gradium TTS (pre-2026.08)155453.7259
11MiniMax Speech 2.8 HD154554.2253
12OpenAudio S2153952.2322
13Lightning v3.1 Pro154453.6295
14Deepdub eTTS 3.2153252.1217
15Eleven Multilingual v2150754.0968
16Eleven Turbo v2.5150553.5951
17MiniMax Speech 2.6 Turbo152550270
18Eleven Flash v2.5150353.2964
19MiniMax Speech 02 HD150353.71028
20Eleven v3152149.3288
21MiniMax Speech 2.6 HD150447.5284
22MiniMax Speech 2.8 Turbo150248.2282
23MiniMax Speech 02 Turbo148350.91037
24Parmesan148152.4984
25Kokoro v1.0147048.8903

Note: Rank/order as served by the site API; as_of is the Space's last modified date. Includes some preliminary/stealth entries.

Source: TTS-AGI community (Hugging Face) · tts-agi-tts-arena-v2.hf.space · results as of 2026-11-11 · retrieved 2026-11-17

Markdown · More Text-to-speech