SALMONN-7B — benchmark results

Provider: Other. Released 2023-11-13. Access: Open.

Unified ELO 1020 ± 33, rank #1771 of 1776 rated models, from 70 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AudioBench - CoVoST2 TA-EN BLEU36.49Mean Score (%)100
AudioBench - WavCaps QA47.3Mean Score (%)100
AudioBench - AudioCaps QA50.29Mean Score (%)87.5
AudioBench - Clotho AQA57.75Mean Score (%)75
AudioBench - IEMOCAP Gender81.31Mean Score (%)75
AudioBench - IMDA GR Sentence59.77Mean Score (%)75
AudioBench - IMDA GR Dialogue42.73Mean Score (%)62.5
AudioBench - VoxCeleb Accent34.22Mean Score (%)62.5
AudioBench - VoxCeleb Gender88.8Mean Score (%)57.1
AudioBench - GigaSpeech WER10.77WER (%) (lower is better)55.6
AudioBench - IMDA Part 4 ASR WER77.57WER (%) (lower is better)55.6
AudioBench - TED-LIUM 3 WER4.6WER (%) (lower is better)55.6

Interactive version: theaggregate.ai/model?slug=salmonn-7b · How the rankings work · Data refreshed daily, snapshot 2026-07-22.