List your company

Hypothesis-mode word alignment on Buckeye 5h (test)

93.1Reference Match RateCanary-Qwen
71.2676.9382.688.27Jul 7, 2026

Evaluation results

MethodLinks
Canary-QwenModel Type=Speech LLM, Align method=Gradients2026.0793.111.49517.1
Canary-QwenModel Type=Speech LLM, Align method=Self-att.2026.0793.111.42123.9
Phi-4-MMModel Type=Speech LLM, Align method=Gradients2026.0792.811.710219
Phi-4-MMModel Type=Speech LLM, Align method=Self-att.2026.0792.811.78016
Crisper-WhisperModel Type=AED, Align method=Official2026.0792.612.24643.5
Parakeet CTCModel Type=CTC, Align method=Gradients2026.0792.311.911614
Parakeet CTCModel Type=CTC, Align method=Posteriors2026.0792.311.99911.1
Crisper-WhisperModel Type=AED, Align method=Gradients2026.079212.75633.8
Crisper-WhisperModel Type=AED, Align method=Cross-att.2026.079212.74459.6
Parakeet TDTModel Type=Transd., Align method=Gradients2026.0791.812.81557.8
Parakeet TDTModel Type=Transd., Align method=Posteriors2026.0791.812.88617.1
Parakeet RNN-TModel Type=Transd., Align method=Gradients2026.0789.813.61567.6
Parakeet RNN-TModel Type=Transd., Align method=Posteriors2026.0789.813.69213.2
OWSM-CTCModel Type=CTC, Align method=Gradients2026.0789.214.31429.9
OWSM-CTCModel Type=CTC, Align method=Posteriors2026.0789.214.31145.4
VoxtralModel Type=Speech LLM, Align method=Gradients2026.0788.315.38725.3
VoxtralModel Type=Speech LLM, Align method=Self-att.2026.0788.315.36737.4
Whisper-large-v3Model Type=AED, Align method=Gradients2026.0788.115.36640.4
Whisper-large-v3Model Type=AED, Align method=Cross-att.2026.0788.115.36640.4
FastConformer (streaming)Model Type=Transd., Align method=Gradients2026.0787.115.92065.9
FastConformer (streaming)Model Type=Transd., Align method=Posteriors2026.0787.115.93180.4
FastConformer (streaming)Model Type=CTC, Align method=Gradients2026.0786.516.81649.7
FastConformer (streaming)Model Type=CTC, Align method=Posteriors2026.0786.516.83630.2
OWLS-1BModel Type=AED, Align method=Gradients2026.0785.720.81,0243.8
OWLS-1BModel Type=AED, Align method=Cross-att.2026.0785.720.81,3127.8
Whisper-baseModel Type=AED, Align method=Gradients2026.0785.5188525.5
Whisper-baseModel Type=AED, Align method=Cross-att.2026.0785.5187735.8
Emformer (streaming)Model Type=Transd., Align method=Gradients2026.0772.130.51557.1
Emformer (streaming)Model Type=Transd., Align method=Posteriors2026.0772.130.53910