A Sound Generation Method Based on Underwater Target and Environmental Information Features

An underwater target and environmental information technology, applied in speech synthesis, neural learning methods, speech analysis, etc., can solve the problems of application limitation of TTS sound generation model and poor underwater acoustic signal effect, so as to improve the effect and improve the feature accuracy. Effect

Active Publication Date: 2022-03-22
HARBIN ENG UNIV
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to solve the problem that the effect of the generated underwater acoustic signal will be poor when the underwater target sound signal feature dictionary and the underwater environment sound signal feature dictionary constructed by the traditional feature extraction method are used to generate the underwater acoustic signal, and the existing The application of the TTS sound generation model in the generation of underwater acoustic signals is limited, and a sound generation method based on underwater targets and environmental information features is proposed.

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 more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A Sound Generation Method Based on Underwater Target and Environmental Information Features
  • A Sound Generation Method Based on Underwater Target and Environmental Information Features
  • A Sound Generation Method Based on Underwater Target and Environmental Information Features

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0044] Specific embodiment one: a kind of sound generation method based on underwater target and environmental information feature described in this embodiment, described method specifically comprises the following steps:

[0045] Step 1. For the underwater target S1, after collecting a sound signal sample of the underwater target (the sound signal sample includes the underwater target signal and the noise signal), the collected sound signal samples are processed in parallel according to the frequency channel, and a system based on Auditory saliency maps for frequency channel processing;

[0046] Step 2. Framing the collected sound signal samples and the constructed auditory saliency map in the time domain to obtain multiple groups of sound signals and auditory saliency maps, and the time lengths of the sound signals and auditory saliency maps in each group are the same;

[0047] The sound signal samples and the auditory saliency map are divided into frames from the beginning ...

specific Embodiment approach 2

[0076] Specific implementation mode two: as figure 1 shown. The difference between this embodiment and the specific embodiment 1 is that in the first step, the collected sound signal samples are processed in parallel according to frequency channels, and an auditory saliency map based on frequency channel processing is constructed. The specific process is as follows:

[0077]The collected sound signal samples are convolved with a bandpass filter formed by superimposing 64 Gammatone filters to obtain the sound signal responses of 64 frequency channels;

[0078] Then the sound signal response of each frequency channel is convoluted in any direction through 8 one-dimensional Gaussian smoothing filters to obtain the convolution result; the convolution result is down-sampled to obtain the sound signal response of each frequency channel in 8 Representation F on a scale i ,i=1,2,…,8, then use F i Compute the auditory salience of the sound signal response for each frequency channel ...

specific Embodiment approach 3

[0088] Specific embodiment three: the difference between this embodiment and specific embodiment two is that: the auditory saliency of each frequency channel sound signal response after normalization on different scales is integrated across scales, and a frequency channel-based The processed auditory saliency map, the specific process is:

[0089]

[0090] In the formula, Map j 'Represents the auditory salience of the sound signal response of the jth frequency channel after cross-scale integration, Map j Represents the auditory salience of the jth frequency channel sound signal response on different scales after normalization, DoG represents a one-dimensional difference Gaussian filter with a length of 50ms, Represents convolution, M nor，j Represents the auditory significance of the jth frequency channel sound signal response on different scales before normalization, j=1,2,...,64;

[0091] Then to Map j ' Perform linear integration to obtain the auditory saliency map A...

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

No PUM Login to view more

Abstract

The invention discloses a sound generation method based on underwater target and environmental information characteristics, which belongs to the research field of underwater acoustic signal generation. The present invention solves the problem that the effect of the generated underwater acoustic signal will be poor when the underwater target sound signal feature dictionary and the environmental sound signal feature dictionary constructed by the traditional feature extraction method are used to generate the underwater acoustic signal, and the existing TTS sound generation model is ineffective. The application of underwater acoustic signal generation is limited. The present invention combines the characteristics of the auditory attention mechanism to highlight the features when extracting the features of the underwater target sound signal and the environmental sound signal, thereby improving the feature accuracy of the underwater target sound signal and the feature dictionary of the environmental sound signal. The feature dictionary is used as the pronunciation dictionary of the sound generation model, embedded in the sound generation model, and the effect of the generated underwater acoustic signal is improved. The invention extends the application field of TTS from the generation of human voice to the generation of underwater acoustic signal. The method of the invention can be applied to the generation of underwater acoustic signals.

Description

technical field [0001] The invention belongs to the research field of generation of underwater acoustic signals, and in particular relates to a sound generation method based on characteristics of underwater targets and environmental information. Background technique [0002] Although the current speech feature extraction methods can more or less achieve good results in speech and audio feature extraction, their audio data sets are all speech, that is, the speech audio of people speaking, distinguishing the speech of people speaking from other Non-speech background noise is relatively easy to do. But for the underwater acoustic signal, this sound is not human voice. This underwater acoustic signal may come from the sound of the propeller of the ship, or from the sound of a motor or environmental noise. Traditional speech feature extraction methods are difficult to distinguish the characteristics of underwater target radiation noise and other noises during feature extraction....

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
no application Login to view more
Patent Type & Authority Patents(China)
IPC IPC(8): G10L13/047G10L13/08G10L25/30G06N3/04G06N3/08
CPCG10L13/047G10L13/08G10L25/30G06N3/08G06N3/045
Inventor 王红滨沙忠澄何鸣王念滨周连科张毅何茜茜
Owner HARBIN ENG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap