List your company

Multi-speaker Automatic Speech Recognition on AMI IHM (test)

16.4cpWERDM-ASR (S2SND) (CN+EN 2900h)
15.406822.110928.81535.5191Apr 24, 2026

Evaluation results

MethodLinks
DM-ASR (S2SND) (CN+EN 2900h)Parms=1.7B, Label perturbation=None, Diarization source=diarization-provided speakers, Time prediction=LLM-predicted times2026.0416.411.4816.934.89
DM-ASR (DiariZen) (CN+EN 2900h)Parms=1.7B, Label perturbation=None, Diarization source=diarization-provided speakers, Time prediction=LLM-predicted times2026.0417.1627.8718.09—
DM-ASR (DiariZen) (EN 1600h)Parms=0.6B, Label perturbation=None, Diarization source=diarization-provided speakers, Time prediction=LLM-predicted times2026.0417.3427.8718.28—
DM-ASR (DiariZen) (CN+EN 2900h)Parms=0.6B, Label perturbation=None, Diarization source=diarization-provided speakers, Time prediction=LLM-predicted times2026.0417.4727.8718.41—
DM-ASR (DiariZen) (EN 630h)Parms=0.6B, Label perturbation=None, Diarization source=diarization-provided speakers, Time prediction=LLM-predicted times2026.0418.4527.8719.52—
VibeVoice-ASRParms=7B2026.0420.4111.9220.82—
Gemini-3.0-ProParms=-2026.0422.3446.2363.65—
JEDIS-LLM(10000h)Parms=5.6B2026.0423.13———
Gemini-2.5-ProParms=-2026.0429.5723.5438.35—
DiariZen + Whisper-large-v3Parms=1.5B2026.0432.2727.6733.99—
Tagspeech(65h)Parms=7B2026.0433.8422.13——
Qwen2.5-Omni-7BParms=7B2026.0441.2337.42——