Method and device for detecting voice quality of hearing aid

Abstract

The invention relates to the technical field of voice quality evaluation, in particular to a method and equipment for detecting the voice quality of a hearing aid. The method comprises the following steps: receiving voice information transmitted by the hearing aid in a noise test environment; voice information is used as input of the self-encoding network, hidden variables in a hidden layer are grouped and replaced, and reconstructed voice information is output to train the self-encoding network; judging whether the reconstructed voice information is normal voice or distorted voice according to the similarity of the feature tensors corresponding to the hidden variables before and after grouping replacement; taking the voice information as the input of a classification network, taking the judgment result of the corresponding reconstructed voice information as a label, outputting the category of the voice information and the confidence coefficient that the voice information is normal voice, and training the classification network; and inputting the received to-be-detected voice transmitted by the hearing aid into the trained classification network, and evaluating the voice quality of the hearing aid according to the output. According to the embodiment of the invention, the voice quality of the hearing aid can be accurately detected.

Description

technical field [0001] The invention relates to the technical field of speech quality evaluation, in particular to a method and equipment for detecting the speech quality of a hearing aid. Background technique [0002] A hearing aid is a small loudspeaker that amplifies the originally inaudible sound, and then uses the residual hearing of the hearing-impaired to send the sound to the auditory center of the brain so that the sound can be felt. The voice quality of hearing aids is one of the most important indicators for evaluating the quality of hearing aids. Generally, voice quality evaluation methods are divided into subjective evaluation and objective evaluation. The subjective evaluation mainly compares the original voice signal and the distorted voice signal by the listener, and scores the voice. Objective evaluation mainly obtains quality evaluation indicators through algorithms to evaluate voice quality, and signal-to-noise ratio is a conventional method to measure vo...

AI Technical Summary

Problems solved by technology

[0003] The subjective evaluation method has the subjectivity of the listener, there may be errors, and the detection results are not accurate enough; while the signal-to-noise ratio algorithm of the objective evaluation often requires a pure voice source signal to be calculated, which is almost impossible in the actual detection process. There must be loss in the process of voice transmission, which makes the evaluation result inaccurate

Images