论文部分内容阅读
本文把成功应用于DNA检验的证据评估方法(似然比)应用于法庭语音证据评估之中,提取语音的LPC作为识别特征,并使用45人电话对话录音中元音/a/作为样本进行了测试。结果表明该方法不仅能正确识别说话人,而且能根据当前嫌疑人样本和问题语音样本的差异,量化该语音样本作为证据的力度,为法庭提供科学合理的证据评估结果和科学解释。同时,自动特征提取的引入比起人工提取共振峰特征,提高了工作效率,识别系统性能也获得大幅提升。
In this paper, the evidence evaluation method (likelihood ratio) successfully applied to DNA test was applied to the assessment of forensic speech evidence, the LPC of speech was extracted as recognition feature, and the vowel / a / test. The results show that this method not only can correctly identify the speaker, but also can quantify the speech sample as evidence according to the difference between the current suspect sample and the question speech sample, and provide the court with scientific and reasonable evidence assessment results and scientific explanations. At the same time, the introduction of automatic feature extraction improves the working efficiency and the performance of the recognition system is greatly improved compared with the feature of formant extraction.