| 雷艺璟,覃晓婧,曹洪林.青年男性普通话与重庆话长时基频特征的时长阈限评估*[J].,2025,44(4):824-833 |
| 青年男性普通话与重庆话长时基频特征的时长阈限评估* |
| Evaluation on the stabilization time of long-term fundamental frequency estimates in Mandarin and Chongqing dialect of young men |
| 投稿时间：2024-09-10 修订日期：2025-06-27 |
| 中文摘要: |
| 长时基频(LTF0)是语音同一性鉴定中最常用的声学特征之一，针对需要多长语料才能得到稳定的LTF0数据的问题，现有行业标准并未做出明确说明。现有研究存在被试人数过少、语料时长较短、说话方式单一等局限性，其研究结果不能直接准确应用于汉语普通话及众多汉语方言。该文基于86位重庆籍青年男性被试在2种语言类型(普通话和重庆话)和2种说话方式(朗读和自由谈话)下的通话录音，采用“稳定值±1%~±5%”5种情况作为LTF0均值的动态阈限标准，对普通话和重庆话的LTF0时长阈限进行了量化分析；为消除语音内容的影响，所有计算经随机化处理并重复100次。结果发现：在“稳定值±1%~±5%”的动态阈限标准下，普通话朗读和自由谈话的平均时长阈限变化范围为2.5~43.3 s，重庆话朗读和自由谈话的平均时长阈限变化范围为3.7~57.9 s；说话方式对时长阈限的影响随着动态阈限范围的扩大而变小；当阈限标准在“稳定值±1%~±2%”时，语言类型和说话方式对时长阈限产生交互效应。结果表明：对于朗读和自由谈话语音，重庆话LTF0的时长阈限均比普通话的更长；相比说话方式，语言类型对LTF0时长阈限的影响更大。 |
| 英文摘要: |
| Long-term fundamental frequency (LTF0) is one of the most commonly used acoustic features in forensic voice comparison. However, the current forensic guidelines do not explicitly specify the minimum duration of speech required to obtain stable LTF0 values. Existing studies are limited by factors such as a small number of participants, short duration of speech samples, and a lack of diversity in speaking styles. As a result, their findings cannot be directly or accurately applied to Standard Chinese or other Chinese dialects. This paper presented a quantitative analysis of the stabilization time of LTF0 in Standard Chinese and Chongqing dialect, based on recordings of 86 young male speakers from Chongqing in two languages (Standard Chinese and Chongqing dialect) and two speaking styles (read and spontaneous speech). Five dynamic threshold criteria ranging from “stable value ±1%” to “stable value ±5%” were used to determine the stability region of the mean values of LTF0. To eliminate the effect of speech content, all calculations were randomized and repeated 100 times. The results showed that under these dynamic threshold criteria, average stabilization times for Standard Chinese LTF0 mean value range from 2.5 s to 43.3 s, while for Chongqing dialect, the range was 3.7 s to 57.9 s. The effect of speaking style on the stabilization time became smaller as the dynamic threshold range expanded. When the threshold criteria were set at from “stable value ±1%” to “stable value ±2%”, an interaction effect between languages and speaking styles was observed on the stabilization time. Languages (Standard Chinese vs. Chongqing dialect) have a greater impact on the stabilization time of LTF0 than speaking styles in forensic voice comparison. Regardless of the speaking styles, the stabilization times for LTF0 are consistently longer in Chongqing dialect than in Standard Chinese. |
| DOI：10.11684/j.issn.1000-310X.2025.04.002 |
| 中文关键词: 语音同一性鉴定 长时基频 时长阈限 普通话 重庆话 |
| 英文关键词: Forensic voice comparison Long-term fundamental frequency Stabilization time Standard Chinese Chongqing dialect |
| 基金项目: |
|
| 摘要点击次数: 1523 |
| 全文下载次数: 1178 |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
