Unverified50% confidenceBenchmarkExact time
The paper validates on the LibriSpeech dataset, covering two different SpeechLLMs, with the maximum bias word list size reaching 2,000 words.
1
Sources
50%
Confidence
Long-term
Relevance
9/25/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedPTC-Bias current experiments are primarily based on the English LibriSpeech dataset and specific SpeechLLMs; its performance in other languages such as Chinese, real noisy environments, and larger bias word lists remains to be validated.65% similarUnverifiedBenchmark datasets for sign language translation (such as PHOENIX-2014T, CSL-Daily, and How2Sign) typically contain only thousands to tens of thousands of samples, far smaller than the millions of sentence pairs in parallel corpora in the machine translation field.65% similarUnverified彭内贝克开发的'语言探究与词语计数'（LIWC）工具通过分析文本语言特征量化心理健康变化，后来成为计算社会科学的重要方法64% similarUnverifiedThe longer the text, the higher the confidence of watermark detection; short texts have high statistical noise and a greater risk of false positives, while detection accuracy is more reliable for texts of several hundred tokens or more.63% similarUnverifiedWhen asked to write a 5,000-word article, GLM 5.2 actually produced more than 5,000 words, with long-form generation capability far exceeding that of most models.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/949145API
curl https://kongchang.com/api/v1/knowledge/claims/949145MCP
get_claim(id=949145)