ModelRadar
/
← All benchmarks

TTS Arena

audio-tts · human-preference↗ Quelle
63Evidence
not yet measured

What it measures

Menschlicher Blindvergleich zweier Sprachsynthesen, Elo-Wertung.

What it does not measure

Latenz, Stimmklonen, Sprachabdeckung.

How the score is composed

Currency50
Discrimination50
Headroom50
Contamination safety100
Gaming resistance100
Methodological rigor80
Reproducibility20

Details

Benchmaxxing risk
low
Difficulty
medium
Public test set
no
Human-verified
yes
LLM-judged
no
Skala
800–1600 Elo (höher besser)

Hand-maintained — no automatic adapter yet.

Best measured models

No results ingested yet.