Unverified50% confidenceFactExact time
Word Error Rate (WER) is the most commonly used evaluation metric in the ASR field; it is calculated by aligning recognition results with reference text and dividing the total number of substitution, deletion, and insertion errors by the number of words in the reference text.
1
Sources
50%
Confidence
Long-term
Relevance
10/7/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedCascaded ASR-based endpoint detection can obtain semantic information, but introduces a strong dependency on transcription results, bringing problems of error propagation and latency accumulation.74% similarUnverified上下文感知解析层通常采用AST分析、依赖图遍历和语义嵌入等技术将底层错误信息转化为高层语义描述67% similarVerifiedWord Error Rate (WER) is calculated as (substitution errors + deletion errors + insertion errors) divided by the total number of reference words.62% similarUnverifiedIn traditional document extraction pipelines, errors from the parsing stage are propagated layer by layer to the extraction stage, affecting the quality of the final results.59% similarVerifiedThe key to error analysis is pinpointing exactly which stage and which type of input caused the failure, rather than blindly adjusting prompts.58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/985824API
curl https://kongchang.com/api/v1/knowledge/claims/985824MCP
get_claim(id=985824)