Unverified50% confidenceBenchmarkExact time
On both the FLEURS and Common Voice datasets, ufakzeka-kulak's WER is lower than Whisper base and Whisper small, and the paired confidence intervals are entirely below zero, so the performance advantage is statistically significant.
1
Sources
50%
Confidence
Long-term
Relevance
11/3/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedWhisper 提供从 tiny 到 large 的多个规格版本,在准确率与计算资源消耗之间提供不同权衡72% similarUnverifiedFor clear English speech, the WER of the Whisper large model can be as low as below 5%; for Chinese, due to word segmentation and homophone issues, WER is usually higher.71% similarUnverifiedThe standard Whisper model experiences reduced accuracy when processing singing, and the community has developed variants such as whisper-timestamped to achieve word-level timeline alignment.70% similarUnverifiedThe word error rate of OpenAI Whisper, as a mainstream ASR engine today, has dropped to below 5%.70% similarUnverifiedWhisper has significantly improved the accuracy of speech-to-text conversion while lowering the barrier to deployment.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1130254API
curl https://kongchang.com/api/v1/knowledge/claims/1130254MCP
get_claim(id=1130254)