List your company
ResearchFormerly WizWand

Automatic Speech Recognition on MLC-SLM (dev)

10.99WER/CERWhisper
9.649618.697327.74536.7927Jan 2, 2026Jan 28, 2026Feb 23, 2026Mar 22, 2026Apr 17, 2026May 13, 2026Jun 9, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Whispertuning=Full-fine-tuned2026.0110.99
Whispertuning=LoRA-fine-tuned2026.0111.4
NTU-Speechlab2026.0111.57
Proposed Speech-LLM2026.0111.74
Seewoo2026.0112.73
SHNU-mASR2026.0113.39
Proposed (MoE + modified CIF)Training data duration (h)=15002026.0615.27
Proposed (MoE + modified CIF)Training data duration (h)=80002026.0615.45
LLM-ASR + MoE ProjectorTraining data duration (h)=15002026.0616.1
LLM-ASR + CIF DownsamplerTraining data duration (h)=15002026.0618.95
Whisper-large-v32026.0621.48
mHuBERT-CTC (baseline)Params(M)=97, TrainingData=MLC-500h, language-known=false, inference_type=single2026.0122.5
LLM-ASR BaselineTraining data duration (h)=15002026.0623.26
mHuBERT-CTC-LIDLoRAParams(M)=107, TrainingData=MLC-500h, language-known=true, inference_type=single2026.0124.6
mHuBERT-CTC-HLoRAParams(M)=102, TrainingData=MLC-500h, language-known=true, inference_type=single2026.0126
mHuBERT-CTC-HLoRAParams(M)=102, TrainingData=MLC-500h, language-known=false, inference_type=single2026.0126.3
mHuBERT-CTC-LIDLoRAParams(M)=107, TrainingData=MLC-500h, language-known=false, inference_type=double2026.0126.6
mHuBERTmode=CTC2026.0129.67
Whisper-smallParams(M)=244, TrainingData=680,000h, language-known=false, inference_type=single2026.0129.7
Whisper-baseParams(M)=74, TrainingData=680,000h, language-known=false, inference_type=single2026.0137.2
mHuBERT-CTC (baseline)Params(M)=97, TrainingData=MSR-1500h, language-known=false, inference_type=single2026.0144.5