The Hearing-Aid Speech Quality Index (HASQI) Version 2
The Hearing-Aid Speech Quality Index (HASQI) Version 2
复制标题
DOI：
10.17743/jaes.2014.0006
复制
发表时间：
2014-03-01
影响因子：
1.4
通讯作者：
Arehart, Kathryn H.
中科院分区：
文献类型：
--
作者：
Kates, James M.;Arehart, Kathryn H.
本文提出了一种改进的语音质量预测指标。助听器语音质量指数(HASQI)版本2是基于听觉外周的模型，该模型包含了由于听力损失而引起的变化。听觉模型输出用于测量由助听器或音频系统处理引起的信号包络、时间精细结构和频谱的变化。该指标包括对噪声和非线性失真敏感的非线性项和对长期光谱变化敏感的线性项，这两个项的乘积构成了质量预测。非线性项结合了包络和时间精细结构修正的测量，与仅基于包络或时间精细结构测量的非线性项相比，它被证明能够产生更准确的预测。与前一版本的HASQI相比，新的指数为各种退化的信号条件提供了更高的精度，包括频率压缩、噪声抑制、用噪声声码器的输出替换语音、声反馈和反馈抵消以及语音与调制噪声的组合。
This paper presents an improved index for predicting speech quality. The Hearing-Aid Speech Quality Index (HASQI) version 2 is based on a model of the auditory periphery that incorporates changes due to hearing loss. The auditory model outputs are used to measure changes in the signal envelope, temporal fine structure, and spectrum caused by the hearing aid or audio system processing. The index comprises a nonlinear term sensitive to noise and nonlinear distortion and a linear term sensitive to long-term spectral changes, and the quality prediction is formed by taking the product of these two terms. The nonlinear term combines measurements of envelope and temporal fine structure modifications, and it is shown to lead to more accurate predictions than using nonlinear terms that are based on envelope or temporal fine structure measurements alone. Compared to the previous version of HASQI, the new index offers improved accuracy for a variety of degraded signal conditions including frequency compression, noise suppression, speech replaced with the output of a noise vocoder, acoustic feedback and feedback cancellation, and speech combined with modulated noise.