蛋鸡发声音频数据库的构建与应用

    Development and application of audio database for laying hens

    • 摘要: 蛋鸡发声含有丰富的机体信息,充分挖掘其声学特性,并利用其无接触、无应激的优点,为建立基于发声信息的蛋鸡养殖远程监测平台提供基础依据。该研究借助音频数字化处理技术和数据库管理平台,以海兰褐蛋鸡为例,搭建系统分别采集其在小规模(5只)饲养条件下的叫声信息及其体态行为。运用音频处理软件Adobe Audition v1.0和音频分析软件Praat 5.3提取蛋鸡发声特征参数,包括持续时间、基音频率、频谱质心、共振峰及其衍生的统计值,以此构建出蛋鸡发声音频数据库,在此基础上分别选取蛋鸡产蛋行为发声、鸣唱声和鸣叫声等典型发声行为对比分析。结果表明,蛋鸡产蛋行为发声与鸣唱声均为多次重复的、有节奏的、短促的音节所构成(称其为句子),前者先抑后扬、后者先扬后抑,句子的音节个数分别是7.8±2.0、15.2±7.7,但其时频域特征间存在着显著差异(P<0.05),与鸣叫声相比,其发声特征参数如频谱质心、共振峰等有着显著差异。研究表明,掌握蛋鸡发声的含义,有助于了解其行为特性、机体状态以及种群间的信息传递,并为蛋鸡行为特征识别与数字化监测平台的构建提供数据支持。

       

      Abstract: Particular states of mood or emotion of laying hens may be accompanied by specific behaviors, vocalization being one of them. Taking advantages of non-invasive monitoring of sound analysis, relevant acoustic features and modern techniques will promote to establish a remote-monitoring platform for laying hens' management. State-of-art digitalization tools and database management technologies were used to the research of behavioral observation and audio analysis. The animals used in this experiment were 5 laying hens (Hy-Line Variety Brown, U.S. standard) and were housed in a feed anticipating condition. During 28 April-30 April 2012, the vocalizations of laying hens were recorded in early morning. A NI-PXI platform integrated with sound sensors were used for all the recordings. The acoustic features including duration of call, pitch, energy center of call, formants and the derivatives were derived using Adobe Audition v1.0 and Praat 5.3. All the features were calculated and written in audio database of laying hens (ADLH). In terms of egg laying, song and call of laying hens, the vocalizations were formed with repeated transient syllables at 7.8±2.0, 15.2±7.7 respectively. Significant differences (P<0.05) were found between duration of call, maximum and minimum frequencies of the verse. With reference to the syllable structure of the vocalizations of laying hens, the acoustic features including energy center, the first formant and the second formant were significantly differed. Pitch of the calls was in the range of 400-500 Hz. Occurrence of the song of laying hens was determined by their emotion and behaviors, and was characterized by walking (43.1%), searching (48.6%), and dust bathing (8.3%). Results indicate that analysis of vocalizations could be an indicator of different physiological conditions and behavior contexts of laying hens and may be an important tool for understanding communications among them. This study will offer an acoustic solution, automatic recognition and digital monitoring tools for laying hens' management.

       

    /

    返回文章
    返回