Unverified50% confidenceBenchmarkExact time
Over the course of one week, six distinct callers hit the Whisper endpoint; Whisper was the largest consumer of GPU time, with the self-built assistant core accounting for 4,970 seconds via the Whisper proxy.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedWhisperFlow achieved a real-world voice input speed of 123 words per minute and can be used anywhere keyboard input is possible, including email, Slack, iMessage, and ChatGPT.73% similarUnverified社区在Whisper基础上开发了whisper-timestamped等变体，能够实现逐词级别的时间轴对齐72% similarVerifiedWhisper基于Transformer编码器-解码器架构，在68万小时的多语言标注音频数据上进行弱监督训练71% similarUnverified标准Whisper模型在处理歌声时准确率会下降，社区开发了whisper-timestamped等变体实现逐词级别时间轴对齐66% similarUnverified围绕 Whisper 构建了 faster-whisper、whisper.cpp 等第三方工具生态66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1002819API
curl https://kongchang.com/api/v1/knowledge/claims/1002819MCP
get_claim(id=1002819)