Acoustic Transducers for Automatic Speech Recognition

用于自动语音识别的声学传感器

• 批准号: 0824271
• 负责人: Eun Kim
• 金额: $ 33.06万
• 依托单位: University of Southern California
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2008
• 资助国家: 美国
• 起止时间: 2008-08-01 至 2012-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0824271&HistoricalAwards=false
• 关键词: Acoustic; Transducers; Automatic; Speech; Recognition

Acoustic Transducers for Automatic Speech RecognitionObjectiveThe research objective is to transform automatic speech recognition in noisy environments by active noise cancellation. The approach is integrate into one package (1) a bank of acoustically filtered microphones that are based on a high quality-factor resonance of a microphone diaphragm and (2) a microspeaker on a silicon chip to improve the signal-to-noise ratio through active noise cancellation. Intellectual MeritThis research will lead to a microspeaker on a silicon chip for sound generation next to a microphone to actively cancel noise, and thus to discriminate noise from speech. One approach for the microspeaker will be to use an elastically compliant polymer layer to form a cantilever-like diaphragm. This research will also develop a resonant microphone with quality factor of about 40 with resonant frequencies that can electrically be tuned. A bank of the resonant microphones with extremely low noise floor will render bandpass-filtered signals at the microphone level, which will be accurate representations of acoustic powers at those specific frequencies.Broader ImpactThe research will greatly impact voice command systems (as well as cellular phones) in subways, airports, running automobiles, battlefields, etc. The tunable resonant microphone and the microspeaker will find broad applications to human-machine interfaces. The proposed activity will also include (1) incorporation of the research results into a lab course on acoustic microelectromechanical systems and (2) research involvement of students from underrepresented groups.

用于自动语音识别的声学换能器目的研究目的通过有源噪声消除来改变噪声环境中的自动语音识别。该方法被集成到一个封装中(1)基于麦克风振膜的高品质因数谐振的一组经过声滤波的麦克风和(2)硅芯片上的麦克风，通过有源噪声消除来提高信噪比。智能价值这项研究将导致在硅芯片上的微型扬声器在麦克风旁边发声，以主动消除噪音，从而区分噪音和语音。微音箱的一种方法是使用弹性顺应性聚合物层来形成悬臂式振膜。这项研究还将开发一种品质因数约为40的谐振麦克风，其谐振频率可以电调。一组噪声极低的共振式麦克风将在麦克风级提供带通滤波信号，这将是特定频率下声功率的准确表示。广泛影响这项研究将对地铁、机场、行驶中的汽车、战场等领域的语音指挥系统(以及手机)产生巨大影响。可调谐谐振式麦克风和麦克风将在人机界面中得到广泛应用。拟议的活动还将包括(1)将研究成果纳入关于声学微电子机械系统的实验室课程，以及(2)让代表人数不足的群体的学生参与研究。