Doherty E T, Shipp T
Speech Research Laboratory, VA Medical Center, San Francisco, CA 94121.
J Speech Hear Res. 1988 Sep;31(3):485-90. doi: 10.1044/jshr.3103.485.
To test for possible contamination of acoustic analyses by record/reproduce systems, five sine waves of fixed frequency and amplitude were sampled directly by a computer and recorded simultaneously on four different tape formats (audio and FM reel-to-reel, audio cassette, and video cassette using pulse code modulation). Recordings were digitized on playback and with the direct samples analyzed for fundamental frequency, amplitude, jitter, and shimmer using a zero crossing interpolation scheme. Distortion introduced by any of the data acquisition systems is negligible when extracting average fundamental frequency or average amplitude. For jitter and shimmer estimation, direct sampling or the use of a video cassette recorder with pulse code modulation are clearly superior. FM recorders, although not quite as accurate, provide a satisfactory alternative to those methods. Audio reel-to-reel recordings are marginally adequate for jitter analysis whereas audio cassette recorders can introduce jitter and shimmer values that are greater than some reported values for normal talkers.
为检测记录/重放系统对声学分析可能造成的污染，通过计算机直接采样固定频率和振幅的五个正弦波，并同时记录在四种不同的磁带格式上（音频和调频开盘式磁带、盒式录音带以及采用脉冲编码调制的盒式录像带）。重放时将录音数字化，并使用过零插值方案对直接采样分析基频、振幅、抖动和微光。在提取平均基频或平均振幅时，任何数据采集系统引入的失真均可忽略不计。对于抖动和微光估计，直接采样或使用采用脉冲编码调制的盒式录像机明显更优。调频录音机虽然不够精确，但为那些方法提供了令人满意的替代方案。开盘式录音带记录对于抖动分析勉强够用，而盒式录音带可能会引入大于一些正常讲话者报告值的抖动和微光值。
