Improving speech intelligibility using acoustic source separation with audio frequency

使用音频声源分离提高语音清晰度

• 批准号: EP/H502335/1
• 负责人: Ahmet Kondoz
• 金额: $ 9.37万
• 依托单位: University of Surrey
• 依托单位国家: 英国
• 项目类别: Research Grant
• 财政年份: 2010
• 资助国家: 英国
• 起止时间: 2010 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=EP%2FH502335%2F1
• 关键词: Improving; speech; intelligibility; using; acoustic

Isolating a single target voice in a noisy environment is a challenging problem, especially forpeople with some degree of hearing loss. This project employs a novel solution that, for thefirst time, utilizes a compact microphone array, works in real-time and in real-life workingconditions with low computational requirements to separate a single voice from a mixture ofvoices and other unwanted sounds. This will improve the intelligibility of speech for peoplewho are hard of hearing with consequent functional and quality of life benefits. Audiofrequency induction-loop systems are the most-widely used assistive technology due to theirearly standardization and simplicity. This project will develop a prototype of an audioprocessing system that can be used immediately with existing induction-loops without theirmodification and will be the first of a series of products for this market.

在嘈杂的环境中分离出单个目标语音是一个具有挑战性的问题，特别是对于有一定程度听力损失的人。该项目采用了一种新颖的解决方案，首次利用紧凑的麦克风阵列，在实时和现实生活中的工作条件下工作，计算要求低，可以将单个语音从语音和其他不需要的声音中分离出来。这将提高听力困难的人的语音清晰度，从而提高功能和生活质量。助听器感应圈系统由于其标准化和简单化的特点，成为目前应用最广泛的辅助技术。该项目将开发一个音频处理系统的原型，该系统可以立即与现有的感应回路一起使用，而无需修改，并将成为该市场的一系列产品中的第一个。