# US 2021/0280204 A1 · Signal processing system and a method of determining noise reduction and compensation thereof

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

- Number: US 2021/0280204 A1
- Document ID: US20210280204A1
- Status: Published application
- Kind: A1
- Application number: 16/881,735
- Filed: 2020-05-22
- Published: 2021-09-09
- Earliest claimed: 2020-03-05
- Assignees: Wistron Corporation
- Applicants: Wistron Corporation
- Inventors: Ching-An Cho; Yu-Yen Chen; Kuo-Ting Huang
- Web page: https://patentgrep.com/patent/US20210280204A1

## CPC

- G10L 21/0232: Processing in the frequency domain
- G06F 17/18: for evaluating statistical data {, e.g. average values, frequency distributions, probability functions, regression analysis (forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04)}
- G10L 13/02: Methods for producing synthetic speech; Speech synthesisers
- G10L 21/02: Speech enhancement, e.g. noise reduction or echo cancellation (reducing echo effects in line transmission systems H04B3/20; echo suppression in hands-free telephones H04M9/08)
- G10L 15/00: Speech recognition (G10L17/00 takes precedence)
- G10L 21/10: Transforming into visible information
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A method of determining noise reduction in a signal includes transforming the signal to generate a spectrogram; determining sharp change in a frequency spectrum for each frame in the spectrogram; and comparing a counted number of frames having sharp change with a predetermined value. The signal is determined to be subject to noise reduction if the counted number is greater than the predetermined value.

## Claims

### Claim 1

A method of determining noise reduction in a signal, comprising:
(a) transforming the signal to generate a spectrogram;
(b) determining sharp change in a frequency spectrum for each frame in the spectrogram; and
(c) comparing a counted number of frames having sharp change with a predetermined value;
wherein the signal is determined to be subject to noise reduction if the counted number is greater than the predetermined value.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
(d) compensating the signal with noise reduction.

### Claim 3 (depends on 2)

The method of claim 2, wherein the step (d) comprises:
adding a noise signal to the signal with noise reduction.

### Claim 4 (depends on 3)

The method of claim 3, wherein the noise signal comprises an additive white Gaussian noise (AWGN) signal.

### Claim 5 (depends on 1)

The method of claim 1, wherein the step (b) comprises:
determining at least one valley point in the frequency spectrum according to a valley threshold, and determining at least one peak point in the frequency spectrum according to a peak threshold;
pairing adjacent valley point and peak point;
determining a slope of each pair of the valley point and the peak point;
wherein sharp change in the frequency spectrum of the frame is determined when a number of pairs having the slope greater than a predetermined slope is equal to or greater than a predetermined threshold.

### Claim 6 (depends on 5)

The method of claim 5, wherein the step (b) further comprises:
averaging a plurality of smallest energy intensities in the frequency spectrum to be the valley threshold, and averaging a plurality of largest energy intensities in the frequency spectrum to be the peak threshold.

### Claim 7 (depends on 1)

The method of claim 1, further comprising:
transform physical quantity to the signal.

### Claim 8 (depends on 7)

The method of claim 7, further comprising:
processing the signal to determine a section containing the physical quantity, and to discard a section not containing the physical quantity.

### Claim 9 (depends on 1)

The method of claim 1, wherein the signal is a voice signal.

### Claim 10 (depends on 9)

The method of claim 9, wherein a mel frequency is used as a frequency unit in the spectrogram.

### Claim 11

A signal compensation method, comprising:
obtaining a pre-compensation energy histogram according to a signal before compensation, the pre-compensation energy histogram having a first standard deviation;
obtaining an energy histogram after adding a noise signal to the signal according to the signal with the added noise signal, the energy histogram after adding the noise signal having a second standard deviation;
obtaining a percentage increase between the second standard deviation and the first standard deviation; and
adjusting the noise signal to be added to the signal when the percentage increase is not within a predetermined range, until the percentage increase associated with the adjusted noise signal is within the predetermined range.

### Claim 12 (depends on 11)

The method of claim 11, wherein the noise signal comprises an additive white Gaussian noise (AWGN) signal.

### Claim 13 (depends on 11)

The method of claim 11, wherein the signal is a voice signal.

### Claim 14

A signal processing system, comprising:
a transducer that transforms physical quantity to a signal;
a signal processor performing the following steps:
(a) transforming the signal to generate a spectrogram;
(b) determining sharp change in a frequency spectrum for each frame in the spectrogram; and
(c) comparing a counted number of frames having sharp change with a predetermined value, the signal being determined to be subject to noise reduction if the counted number is greater than the predetermined value;
a noise source that provides a noise signal; and
a synthesizer that adds the noise signal and the signal with noise reduction, thereby obtaining a compensated signal.

### Claim 15 (depends on 14)

The system of claim 14, wherein the noise signal comprises an additive white Gaussian noise (AWGN) signal.

### Claim 16 (depends on 14)

The system of claim 14, further comprising:
a recorder that records the signal by performing sampling and analog-to-digital conversion on the signal; and
a storage device that stores the signal recorded by the recorder.

### Claim 17 (depends on 14)

The system of claim 14, wherein the signal is a voice signal.

### Claim 18 (depends on 17)

The system of claim 17, wherein a mel frequency is used as a frequency unit in the spectrogram.
