Voice enhancement algorithm combining cochlear voice features and hopping deep neural network
A deep neural network, voice feature technology, applied in biological neural network models, neural learning methods, voice analysis, etc., to achieve good enhancement effects, improve voice enhancement effects, and clear voice features.
Active Publication Date: 2020-11-06
HARBIN UNIV OF SCI & TECH
View PDF9 Cites 2 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Although MRCG and MRACC have better speech enhancement effects in low SNR environments, there is still room for improvement.
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0060] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.
[0061] In the embodiment of the present invention: the speech enhancement algorithm of joint cochlear speech feature and jump deep neural network, comprises the following steps:
[0062] The first step: use MMSE as the front-end improved MRCG speech feature, and compare the speech enhancement effect of different features as network input;
[0063] Step 2: Analyze the ability of DNN and Skip-DNN to filter out "music noise", and establish a speech enhancement mo...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a voice enhancement algorithm combining cochlear voice features and a hopping deep neural network. According to the voice enhancement algorithm combining cochlear voice features and the hopping deep neural network, improved MRCG speech features with MMSE as the front end are adopted; different features are compared with one another so as to serve as the speech enhancement effect of network input; the ability of DNN and Skip-DNN to filter music noise is analyzed, a voice enhancement model combining an improved MRCG and Skip-DNN is established; and experiments show that the voice enhancement model combining the improved MRCG and Skip-DNN has a better enhancement effect than MRACC and MRCG, and meanwhile, in a low-signal-to-noise-ratio and non-stationary background noise environment, the Skip-DNN filters out part of music noise, so that the complex nonlinear relationship between the MRCG and the IRM is well fitted and improved, the speech enhancement effect in thelow signal-to-noise ratio environment is improved, clearer speech features are obtained, and the speech intelligibility and speech quality can be effectively improved.
Description
technical field [0001] The invention relates to the technical field of speech enhancement, in particular to a speech enhancement algorithm combining cochlear speech features and jumping deep neural networks. Background technique [0002] Single-channel speech enhancement is an interesting and challenging technology whose main purpose is to improve speech quality, enhance speech intelligibility, and make target speech clearer in noisy environments. Because of its more practical functions, it has many applications in engineering, such as hearing aids, communication equipment, and robust speech recognition. Single-channel speech enhancement has played an important role. [0003] For decades, many people have devoted themselves to the research of single-channel speech enhancement and proposed many methods. It can be roughly divided into two types, unsupervised and supervised speech enhancement algorithms. Among them, unsupervised algorithms include spectral subtraction, Wiener...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G10L21/0208G10L21/0232G10L25/30G06N3/08G06N3/04
CPCG10L21/0208G10L21/0232G10L25/30G06N3/08G06N3/045
Inventor 兰朝凤刘春东张磊康守强郭小霞韩闯
Owner HARBIN UNIV OF SCI & TECH