Advanced noise reduction and speech extraction for hearing aids and speech communication devices

从这项研究计划进行的研究是为数字语音处理设备中的降噪和语音提取开发创新和有效的解决方案。特别是，重点将放在助听器上，尽管一些预期的研究成果也适用于其他语音设备，如免提或移动的电话和自动语音识别系统中使用的设备。助听器已经有几十年的历史了，但今天助听器的采用率仍然很低。虽然有几个社会经济因素会影响选择是否使用助听器的决定，但其中一个因素是，在一些具有挑战性但常见的声学环境下，助听器的整体性能仍然相当差，因为降噪和语音提取算法的性能有限。其他重要的应用与助听器具有相同的缺点：例如，移动的电话设备、蓝牙耳机和语音识别系统，其中性能在一些具有挑战性的声学环境下显著降低。尽管多年来在降噪和语音提取领域已经取得了非常显著的工作量，但是仍然存在有待解决的问题和有待改进的算法，以使这些系统在几个困难的实际环境中令人满意地工作。在该项目中进行的新研究提出了创新，以解决限制降噪和语音提取系统性能的一些主要问题。成功完成拟议的研究将大大提高降噪和语音提取技术的整体性能，从而在几个应用中实现切实的改进，这些应用从医疗保健的角度和商业的角度来看都很重要。

The research to be conducted from this research proposal is for the development of innovative and efficient solutions for noise reduction and speech extraction in digital speech processing devices. In particular, the emphasis will be on hearing aids, although several of the expected research outcomes are also applicable to other speech devices such as the ones used in hands-free or mobile telephony and automatic speech recognition systems. Hearing aids have been available for a few decades, but the hearing aid adoption rate still remains low today. While there are several socio-economic factors that affect the decision of opting whether or not to use a hearing aid, one factor is that under some challenging yet common acoustic environments, the overall performance of hearing aids has remained fairly poor because of the limited performance of noise reduction and speech extraction algorithms. Other important applications share the same shortcomings as hearing aids: for example mobile telephony devices, Bluetooth headsets, and speech recognition systems, where the performance degrades significantly under some challenging acoustic environments. Even though a very significant amount of work has been achieved in the area of noise reduction and speech extraction over the years, there are still issues to be solved and algorithms to be improved for those systems to work satisfactorily in several difficult practical environments. The new research to be conducted in this project proposes innovations to address some of the main issues limiting the performance of noise reduction and speech extraction systems. A successful completion of the proposed research would go a long way in improving the overall performance of noise reduction and speech extraction technologies, leading to tangible improvements in several applications which are important both from a healthcare perspective and a commercial perspective.