Unverified50% confidenceFactExact time
Pretrained ASR systems (such as Whisper and Qwen3-ASR) perform well on standard corpora but perform poorly in high-noise domains such as broadcast police communications.
1
Sources
50%
Confidence
Long-term
Relevance
9/28/2026
First Seen
Sources
Related Entities
Related Claims
Unverified紧急报警场景中的语音特征（高语速、气喘、哭泣、语句中断重叠）在主流ASR训练数据集中严重不足，形成显著的领域偏移问题73% similarUnverifiedAudio tokens are trained with the objective of reconstructing high-fidelity audio, retaining all acoustic information including timbre, emotion, and ambient sound, rather than preserving only linguistic content as ASR output does.70% similarUnverified通用ASR模型的训练数据大多来自播客、有声书、YouTube视频等宽带高质量音频源，与呼叫中心的窄带嘈杂环境存在显著差异68% similarUnverifiedIn real-world call recordings with background noise, colloquial expressions, and overlapping speakers, ASR accuracy typically drops below standard benchmarks.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/955549API
curl https://kongchang.com/api/v1/knowledge/claims/955549MCP
get_claim(id=955549)