/

← All benchmarks

FLEURS

audio-stt · multilingual-asr↗ Quelle

52Evidence

saturated

What it measures

Spracherkennung in 102 Sprachen, vergleichbar aufgebaut.

What it does not measure

Spontansprache, Laerm.

How the score is composed

Currency100

Discrimination6

Headroom1

Contamination safety55

Gaming resistance100

Methodological rigor80

Reproducibility60

Details

Benchmaxxing risk: low
Difficulty: medium
Public test set: yes
Human-verified: yes
LLM-judged: no
Letzter Messwert: 2026-09-28
Skala: 0–100 % WER (niedriger besser)

Best measured models

1azure-speech-07Microsoft0.87 % WER
2elevenlabs-scribe-v20.9 % WER
3assemblyai-universal-3-5-pro1.01 % WER
4azure-speech-06Microsoft1.03 % WER
5assemblyai-universal-3-pro1.56 % WER
6azure-speech-06Microsoft1.75 % WER
7azure-speech-07Microsoft1.75 % WER
8azure-speech-07Microsoft1.83 % WER
9elevenlabs-scribe-v21.85 % WER
10Phi-4-multimodalMicrosoft1.88 % WER