Speaker intelligibility detection method for cochlear implant signals in noisy environment
A technology of cochlear implants and detection methods, which is applied in speech analysis, speech recognition, instruments, etc., and can solve problems such as the decline of speech comprehension of cochlear implants
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Publication Date
- 2020-10-20
Smart Images

Figure 1 
Figure 2 
Figure 3
Abstract
Description
technical field
[0001] The invention relates to the field of speech signal processing, in particular to a method for detecting the speech intelligibility of different speakers in a noise environment for the speech signal processed by a cochlear implant. Background technique
[0002] At present, research on cochlear implants mainly focuses on the method of cochlear implants for sound processing, the intelligibility of cochlear implant users to different voices, and the development and improvement of cochlear implants or systems, such as patents CN103892939A, CN101716108A, CN101953722A, etc. Research on sound processing methods from cochlear implants; patent CN101770549A discloses an automatic detection method and detection system for speech processing strategies based on a "virtual processor", which detects the correctness of the implementation of cochlear implant processing algorithms, etc. Most of the current research focuses on the improvement of different processing algor...
Examples
Embodiment Construction
[0069] Include the following steps:
[0070] (1) The cochlear implant treatment process is as follows, such as figure 1 shown;
[0071] (1), preprocessing of speech signal
[0072] The preprocessing flow of the speech signal is as follows: figure 2 As shown, the process includes: input speech signal, endpoint detection, RMS normalization, adjusting the input signal sampling rate to F s , pre-emphasis, framing and windowing. Here, the energy detection method is used for endpoint detection; after the root mean square is normalized, the root mean square value of the input signal is 1; F s Take 16000Hz, if the sampling rate of the signal is not equal to F s , the signal needs to be resampled to F s ; Use a high-pass filter with a cutoff frequency of 1200Hz for pre-emphasis; use a Hamming window for frame division, wherein each frame is 8ms long (L=128 points), and N short-time frame signals with a frame length of L are obtained after preprocessing ;
[0073] (2), carry ou...