Automatic Speech Recognition on SLPHelmUltraSuitePlus (test)
0.194WERWhisper-large-v3-turbo
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Whisper-large-v3-turboCondition=Pure Transcr., Model Category=Fine-tuned ASR2026.04 | 0.194 | 0.64 | 0.814 | 0.708 | 0.187 | 0.365 | |
| Whisper-large-v2Condition=Pure Transcr., Model Category=Fine-tuned ASR2026.04 | 0.26 | 0.607 | 0.793 | 0.666 | 0.228 | 0.21 | |
| Whisper-large-v2Condition=Incl. Vowel Descr., Model Category=Fine-tuned ASR2026.04 | 0.293 | 0.555 | 0.703 | 0.588 | 0.273 | 0.295 | |
| Whisper-large-v3-turboCondition=Incl. Vowel Descr., Model Category=Fine-tuned ASR2026.04 | 0.303 | 0.553 | 0.71 | 0.584 | 0.278 | 0.252 | |
| Whisper-large-v3Condition=Pure Transcr., Model Category=Fine-tuned ASR2026.04 | 0.369 | 0.52 | 0.748 | 0.563 | 0.314 | 0.33 | |
| Whisper-large-v3Condition=Incl. Vowel Descr., Model Category=Fine-tuned ASR2026.04 | 0.495 | 0.414 | 0.596 | 0.456 | 0.4 | 0.392 | |
| Qwen2-Audio-7B-InstructModel Category=LLM SOTA2026.04 | 0.572 | — | — | 0.547 | 0.385 | — | |
| Qwen2.5-Omni-3BModel Category=LLM SOTA2026.04 | 0.996 | — | — | 0.36 | 0.569 | — | |
| Qwen2.5-Omni-7BModel Category=LLM SOTA2026.04 | 1.762 | — | — | 0.44 | 0.489 | — |