# US 12,222,430 B2 · High-resolution sound source map obtaining and analyzing method and system using artificial intelligence neural network

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 12,222,430 B2
- Document ID: US12222430B2
- Status: Granted
- Kind: B2
- Application number: 17/766,476
- Filed: 2021-05-27
- Published: 2023-05-25
- Granted: 2025-02-11
- Earliest claimed: 2020-09-07
- Assignees: Korea Research Institute of Standards and Science; Postech Research and Business Develpment Foundation
- Applicants: Korea Research Institute of Standards and Science; POSTECH Research and Business Development Foundation
- Inventors: Ji Ho Chang; Seung Chul Lee; Soo Young Lee
- Web page: https://patentgrep.com/patent/US12222430B2

## CPC

- G01S 3/8083: {determining direction of source}
- G06T 11/26: Drawing of charts or graphs
- G01S 3/808: using transducers spaced apart and measuring phase or time difference between signals therefrom, i.e. path-difference systems
- G01S 5/0278: {involving statistical or probabilistic considerations (G01S5/0252, G01S5/0294 take precedence)}
- G01S 5/22: Position of source determined by co-ordinating a plurality of position lines defined by path-difference measurements (G01S5/28 takes precedence)
- G06N 3/04: Architecture, e.g. interconnection topology
- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- G01S: RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A method and system for generating a target map for training a neural network and obtaining a sound source map regardless of the maximum number of sound sources, having a short computation time for inference, high spatial resolution and high sound source accuracy. The method includes generating grids each having a spacing within a given range at positions where sound sources are present in order to form a sound source map, calculating a result value for each of coordinates of the grids so that the result value is a local maximum at the position of a sound source and the result value decreases depending on the distance from the sound source, arranging the result values at positions on matrices corresponding to the respective coordinates of the grids, and generating a target map having an image form by using the result values arranged in on the matrices.

## Claims

### Claim 1

A method of obtaining and analyzing a high-resolution sound source map using an artificial intelligence neural network, the method comprising:
a grid generation step of generating grids, via software executing on a processing unit, each having an interval within a given range at locations where sound sources are present, in order to form a computer generated sound source map including a plurality of pixels;
a calculation step of calculating a result value for each of coordinates of the grids, wherein the result value is a maximum at a location of the sound source and the result value is decreased based on a distance from the sound source;
a matrix generation step of arranging the result values at locations on matrices corresponding to the respective coordinates of the grids; and
a target map generation step of generating a target map having an image form by using the result values arranged in the matrix.

### Claim 2 (depends on 1)

The method of claim 1, wherein the result value is defined through Equation 1 below:
f
⁡
(
R
)
=
ε
R
N
+
ε
wherein R means a distance between the sound source and the grid and N and ε mean constants).

### Claim 3 (depends on 1)

The method of claim 1, wherein the target map generation step further comprises, when the sound source is present in plural, a target map synthesis step of generating target maps for the respective sound sources and then synthesizing the target maps as one target map.

### Claim 4 (depends on 1)

The method of claim 1, further comprising a neural network training step of:
designating, as a comparison input value, a beamforming map generated by a sound source having a location and intensity identical with a location and intensity of a sound source used to generate the target map,
designating the target map as a comparison output value, and
training an artificial neural network to output the comparison output value corresponding to the comparison input value when receiving an image identical with an image of the comparison input value.

### Claim 5 (depends on 4)

The method of claim 4, further comprising:
a target map inference step of receiving the beamforming map and inferring and outputting a target map corresponding to the beamforming map by using the artificial neural network; and
an analysis step of analyzing the target map and extracting information of the sound source.

### Claim 6 (depends on 5)

The method of claim 5, wherein the analysis step further comprises:
a sound source location extraction step of extracting, from the target map, a plurality of regions each having a given size and comprising a point at which the result value is a maximum value;
a grid split step of splitting each of the regions into regions each having an interval smaller than that of the grid of the target map; and
a sound source information analysis step of analyzing information on each sound source included in each region.

### Claim 7 (depends on 1)

A non-transitory computer-readable recording medium on which a program for implementing the method according to claim 1 has been recorded.

### Claim 8

A system for obtaining and analyzing a high-resolution sound source map using an artificial intelligence neural network, the system comprising:
a grid generation unit, including software executing on a processor unit, configured to generate grids each having an interval within a given range around sound sources, in order to form a computer generated sound source map including a plurality of pixels;
a result value calculation unit, including software executing on a processor unit, configured to calculate a result value for each coordinate of the grid, wherein a result value is a maximum at a location of the sound source and the result value is decreased based on a distance from the sound source;
a matrix generation unit, including software executing on a processor unit, configured to arrange the result values at locations on matrices corresponding to the respective coordinates of the grids; and
a target map generation unit, including software executing on a processor unit, configured to generate a target map having an image form by using the result values arranged in on the matrices.

### Claim 9 (depends on 8)

The system of claim 8, wherein the result value is defined through Equation 1 below:
f
⁡
(
R
)
=
ε
R
N
+
ε
(wherein R means a distance between the sound source and the grid, and N and ε mean constants).

### Claim 10 (depends on 8)

The system of claim 8, further comprising a target map synthesis unit, wherein, when the sound source is present in plural, the target map synthesis unit generates target maps for the respective sound sources and synthesizes the target maps as one target map.

### Claim 11 (depends on 8)

The system of claim 8, further comprising a neural network training unit configured to designate, as a comparison input value, a beamforming map generated by a sound source having a location and intensity identical with a location and intensity of a sound source used to generate the target map, designate the target map as a comparison output value, and train an artificial neural network to output the comparison output value corresponding to the comparison input value when receiving an image identical with an image of the comparison input value.

### Claim 12 (depends on 11)

The system of claim 11, further comprising:
an inference unit configured to infer and output the beamforming map as the target map corresponding to the beamforming map by using the artificial neural network; and
an analysis unit configured to analyze the target map and extract information of the sound source.

### Claim 13 (depends on 12)

The system of claim 12, wherein the analysis unit further comprises:
a sound source location extraction unit configured to extract, from the target map, a plurality of regions each having a given size and comprising a point at which the result value is a maximum value;
a grid split unit configured to split each of the regions into regions each having an interval smaller than that of the grid of the target map; and
a sound source information analysis unit configured to analyze information for each sound source included in each of the regions.