# US 2024/0176815 A1 · Artificial intelligence-based similar sound source search system and method

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

- Number: US 2024/0176815 A1
- Document ID: US20240176815A1
- Status: Published application
- Kind: A1
- Application number: 18/222,405
- Filed: 2023-07-14
- Published: 2024-05-30
- Earliest claimed: 2022-11-30
- Assignees: POZALabs Co., Ltd.
- Applicants: POZALabs Co., Ltd.
- Inventors: Tae Hyun KIM; Jun Byeong Lee; Hyo Seob KIM; Hee Taek LIM
- Web page: https://patentgrep.com/patent/US20240176815A1

## CPC

- G06F 16/63: Querying
- G06F 16/683: using metadata automatically derived from the content
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/08: Learning methods
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

The present disclosure relates to an artificial intelligence-based similar sound source search system and method for searching for a matching sound source by identifying timbre, pitch, and chord information using a convolutional neural network (CNN) and comprises a sound source input unit, a sound source analysis unit, a sound source information extraction unit, and a matching sound source search unit.

## Claims

### Claim 1

An artificial intelligence-based similar sound source search system comprising:
a sound source input unit for receiving a reference sound source;
a sound source analysis unit for analyzing waveform of the reference sound source input through the sound source input unit;
a sound source information extraction unit for extracting sound source information data including at least one of genre, beats per minute (BPM), scale, and chord progression information for a reference sound source by applying the analysis result of the sound source analyzing unit as an input value of an artificial intelligence model by the artificial intelligence model; and
a matching sound source search unit for searching for a matching sound source that matches a predetermined standard or more by comparing the sound source information data stored in the database with the sound source information data of the reference sound source extracted from the sound source information extraction unit.

### Claim 2 (depends on 1)

The system of claim 1, wherein the sound source analysis unit performs waveform analysis using constant Q transform (CQT) or frequency analysis on the waveform of the sound source.

### Claim 3 (depends on 1)

The system of claim 1, wherein the sound source information extraction unit includes a BPM extraction model, a genre extraction model, a SCALE classification model, and a chord progression extraction model.

### Claim 4 (depends on 3)

The system of claim 3, wherein the BPM extraction model identifies timbre, pitch, and chord information by convolutional neural network (CNN) and derives a BPM value using fully-connected (FC) or other deep learning models.

### Claim 5 (depends on 3)

The system of claim 3, wherein the scale extraction model identifies timbre, pitch, and chord information by convolutional neural network (CNN) and derive a SCALE value using fully-connected (FC) or other deep learning models.

### Claim 6 (depends on 3)

The system of claim 3, wherein the genre extraction model calculates information for each frequency band, performs time series analysis and uses deep learning models including convolutional neural network (CNN) and self-attention to extract correlations between data.

### Claim 7 (depends on 3)

The system of claim 3, wherein the chord progression extraction model derives chord information by extracting the relationship between chords using a deep learning model including bidirectional self-attention.

### Claim 8 (depends on 1)

The system of claim 1, wherein the matching sound source search unit receives a predetermined standard for searching for a matching sound source that matches a predetermined standard or more.

### Claim 9

An artificial intelligence-based similar sound source search method executed on a computer device,
wherein the computer device includes at least one processor configured to execute computer readable instructions contained in a memory,
wherein the artificial intelligence-based similar sound source search method comprises
a sound source input step for receiving a reference sound source;
a sound source analysis step for analyzing waveform of the reference sound source input through the sound source input unit;
a sound source information extraction step for extracting sound source information data including at least one of genre, beats per minute (BPM), scale, and chord progression information for a reference sound source by applying the analysis result of the sound source analyzing unit as an input value of an artificial intelligence model by the artificial intelligence model; and
a matching sound source search step for searching for a matching sound source that matches a predetermined standard or more by comparing the sound source information data stored in the database with the sound source information data of the reference sound source extracted from the sound source information extraction step.

### Claim 10 (depends on 9)

The method of claim 9, wherein the sound source analysis step performs waveform analysis using constant Q transform (CQT) or frequency analysis on the waveform of the sound source.

### Claim 11 (depends on 9)

The method of claim 9, wherein the sound source information extraction step includes a BPM extraction model, a genre extraction model, a scale classification model, and a chord progression extraction model.

### Claim 12 (depends on 9)

The method of claim 9, wherein the matching sound source search step receives a predetermined standard for searching for a matching sound source that matches a predetermined standard or more.
