Unverified50% confidenceBenchmarkExact time
With greedy decoding, Whisper large-v3 achieves a FLEURS Turkish WER of 0.0604 and a Common Voice 27.0 Turkish WER of 0.0767.
1
Sources
50%
Confidence
Long-term
Relevance
11/3/2026
First Seen
Sources
Related Entities
Related Claims
Unverifiedv3.29.0 adds the shida voice for Live speech.64% 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.63% similarUnverifiedOn 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.63% similarUnverifiedWhisper offers multiple sizes from tiny to large, with parameter counts ranging from about 39 million to 1.55 billion, is highly robust to differences in accent, background noise and recording quality, and supports language identification and translation.62% similarUnverifiedSteerTTS-edge's vocoder is a 6.2-million-parameter Vocos-style model distilled from charactr/vocos-mel-24khz, also using int4 QAT.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1130255API
curl https://kongchang.com/api/v1/knowledge/claims/1130255MCP
get_claim(id=1130255)