Title:
SPEECH SIGNAL PROCESSING METHOD, NEURAL NETWORK TRAINING METHOD AND DEVICE
Document Type and Number:
WIPO Patent Application WO/2024/050802
Kind Code:
A1
Abstract:
The present application discloses a speech signal processing method, a neural network training method and a device, which are applied to the field of signal processing. The speech signal processing method comprises: acquiring a first bone conduction speech signal, extracting an excitation parameter from the first bone conduction speech signal, and determining a first transfer function corresponding to the first bone conduction speech signal; then inputting the first transfer function into a trained neural network to obtain a second transfer function (i.e., a transfer function of a predicted air conduction speech signal), the trained neural network being obtained by training a neural network using a training data set on the basis of a target loss function; and finally according to the second transfer function and the excitation parameter, obtaining a first air conduction speech signal corresponding to the first bone conduction speech signal. According to the present application, on the basis of the trained neural network, the transfer function of the bone conduction speech signal is mapped into the transfer function of the air conduction speech signal, so that tone compensation from the bone conduction speech signal to the air conduction speech signal is realized, the listening effect of the bone conduction speech signal is brighter and more comfortable, and the call efficiency is improved.
Inventors:
ZHANG LIBIN (CN)
Application Number:
PCT/CN2022/117989
Publication Date:
March 14, 2024
Filing Date:
September 09, 2022
Export Citation:
Assignee:
HUAWEI TECH CO LTD (CN)
International Classes:
G10L21/0308; G10L21/0208; H04R3/00
Foreign References:
| CN112017687A | 2020-12-01 | |||
| CN112599145A | 2021-04-02 | |||
| CN112565977A | 2021-03-26 |
Attorney, Agent or Firm:
SHENPAT INTELLECTUAL PROPERTY AGENCY (CN)
Download PDF: