摘要
为了探究运动状态下说话人的语音特征,提高语音同一认定的准确性,以0 km/h、5 km/h、10 km/h、15 km/h步速下的语音样本为研究对象,从听觉特征和频谱特征两方面总结出了易受和不易受步速影响的语音特征,以期为司法语音检验提供参考。结果表明,实验者在不同步速下的听觉特征具有明显的差异,如语速、音量、节奏感、流畅度、连贯性、环境噪声、喷麦气流等。在频谱特征方面,单元音“a”“u”“i”“e”的共振峰走向特征相对稳定,二合元音“ue”“ei”的共振峰走向特征受到步速影响较大,“ai”和“ao”受到步速影响较小,三合元音“iao”的共振峰走向特征最为稳定,基本不受步速的影响。在语音同一认定过程中,应尽量使检材和样本中的发音人处于相同的运动状态。在不明确检材中发音人运动状态的情况下,应谨慎使用易受运动影响的语音特征进行同一认定。
To explore phonetic features of speakers in kinestate and improve the accuracy of forensic voice identification,phonetic samples at 0 km/h,5 km/h,10 km/h,and 15 km/h walking speeds are taken as the research objects.The phonetic features that are easily or not easily affected by walking speed are summarized according to auditory and spectral characteristics,which may provide reference for judicial speech testing.The results show that there are obvious differences in auditory characteristics among participants at different walking speeds,such as speech rate,volume,rhythm,fluency,coherence,ambient noise,breathing,pops,etc.In terms of spectral characteristics,the formant trends of the monophthongs“a”“u”“i”and“e”exhibit relative stability.As for diphthongs,“ue”and“ei”are significantly influenced by walking speed,while those of“ai”and“ao”are less affected.The triphthong“iao”is the most stable and is basically not affected by walking speed.During forensic voice identification,the speakers in the samples should be recorded in the same state of motion as those in the reference samples.If motion state cannot be confirmed,the phonetic features that are easily affected by walking speed should be carefully used in identification.
作者
潘洪兵
刘德臣
高海洋
PAN Hongbing;LIU Decheng;GAO Haiyang(Department of Criminal Science and Technology,Jilin Police College,Changchun 130117,China;Criminal Police Detachment,Changchun Public Security Bureau,Changchun 130028,China)
出处
《中国人民公安大学学报(自然科学版)》
2025年第1期24-32,共9页
Journal of People’s Public Security University of China(Science and Technology)
基金
2023年吉林警察学院科研项目(jykyyb202312)
2023年吉林省公安厅科研项目(JSYJ04-2023)
吉林省教育科学规划2024年度一般课题(GH24823)
2023年吉林省教育厅科学技术研究规划项目(JJKH20231084KJ)。
关键词
运动状态
语音特征
语音同一认定
声纹检验
motion state
phonetic features
forensic voice identification
voiceprint recognition
