The Recorder's Paradox

The need to access vernacular speech styles, which appear in informal, uncontrolled environments poses a problem for the recording of audio high enough in quality for acoustic analysis. As a result, sociolinguistic interviews are either relatively formal and controlled resulting in more formal speech and decent quality audio OR informal, uncontrolled contexts with a decent level of vernacular speech yet low quality audio. This poses a problem for the analysis of vernacular speech. Studies show the effects of noise on voice quality properties but are noisy environments actually detrimental to analysis of formant frequencies? Does noise affect all formants equally? What types of noise should be avoided?获取出现于非规范、无约束场景下的通俗言语风格（vernacular speech styles），为采集满足声学分析（acoustic analysis）要求的高音质录音带来了挑战。为此，社会语言学访谈（sociolinguistic interviews）往往陷入两难境地：要么采用相对规范且受约束的形式，产出的言语风格较为正式且音频质量尚可；要么处于非规范、无约束的场景，虽能采集到足量的通俗言语，但音频质量偏低。这为通俗言语的分析研究带来了难题。已有研究证实了噪声对语音质量特征的影响，但嘈杂环境是否真的会对共振峰频率（formant frequencies）分析产生不利影响？噪声是否会对所有共振峰产生同等影响？应当规避何种类型的噪声？