Hearing aid speech enhancement method based on deep domain adaptive network

Abstract

The invention discloses a hearing aid speech enhancement method based on a deep domain adaptive network, comprising: extracting frame-level logarithmic power spectrum features from noisy speech and clean speech respectively; constructing a deep learning model based on an encoder-decoder structure As a baseline speech enhancement model; on the basis of the baseline speech enhancement model, a transfer learning speech enhancement model based on a deep domain adaptive network is constructed; the transfer learning speech enhancement model introduces a domain adaptation layer and Relative discriminator; use domain adversarial loss to train the transfer learning speech enhancement model; in the enhancement stage, according to the trained deep domain adaptive transfer learning speech enhancement model, input the frame-level LPS features of the noisy speech in the target domain, and reconstruct the enhanced speech waveform . The invention encourages a feature encoder to generate domain invariant features through domain confrontational training, thereby improving the adaptability of the speech enhancement model to unseen noise.

Description

technical field [0001] The invention relates to the technical field of speech enhancement, in particular to a hearing aid speech enhancement method based on a deep domain adaptive network. Background technique [0002] In a complex environment, the target sound is usually submerged in the noise, and the results of the sound spectrum analysis are seriously affected, which makes the performance of the adaptive frequency reduction algorithm drop sharply. At the same time, some hearing-impaired characteristics of hearing-impaired patients, such as high hearing threshold, difficulty in identifying short-term features, and degenerated auditory periphery, make speech understanding in complex scenes a common problem and difficult problem that affects the usage rate. [0003] Classical single-channel noise suppressors are based on statistical signal processing methods, which focus on how to effectively estimate the noise spectrum from noisy speech to suppress it. Typical algorithms ...

AI Technical Summary

Problems solved by technology

This patent realizes the migration of unknown noise types and unknown signal-to-noise ratios through the migration learning algorithm, but only realizes the migration of one type of noise to another noise, and its applicable scenarios have limitations

Secondly, the algorithm has limited transfer effect when the noise type and signal-to-noise ratio do not match.

Images