ELECTRONIC DEVICE AND METHOD CAPABLE OF VOICE RECOGNITION

• US 20170084292A1
• Filed: 07/22/2016
• Published: 03/23/2017
• Est. Priority Date: 09/23/2015
• Status: Active Grant

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: YOO, Jong-uk
• Granted Patent: US 10,056,096 B2
• Time in Patent Office: Days
• CPC Class Codes: G10L 2025/783 based on threshold decision
• CPC Class Codes: G10L 25/09 the extracted parameters being zero crossing rates
• CPC Class Codes: G10L 25/18 the extracted parameters being spectral information of each sub-band
• CPC Class Codes: G10L 25/84 for discriminating voice from noise

Abstract

Provided herein is an electronic device and method of voice recognition, the method including analyzing an audio signal of a first frame when the audio signal is input and extracting a first feature value; determining a similarity between the first feature value extracted from the audio signal of the first frame and a first feature value extracted from an audio signal of a previous frame; analyzing the audio signal of the first frame and extracting a second feature value when the similarity is below a predetermined threshold value; and comparing the extracted first feature value and the second feature value and at least one feature value corresponding to a pre-defined voice signal and determining whether or not the audio signal of the first frame is a voice signal, and thus the electronic device may detect only a voice section from the audio signal while improving the processing speed.

Claims

1. A voice recognition method of an electronic device, the method comprising:

analyzing an audio signal of a first frame when the audio signal of the first frame is input into the electronic device using an inputter of the electronic device, and extracting a first feature value using a processor of the electronic device;

determining a similarity between the first feature value extracted from the audio signal of the first frame and a first feature value extracted from an audio signal of a previous frame using the processor;

analyzing the audio signal of the first frame and extracting a second feature value when the similarity is below a predetermined threshold value using the processor; and

comparing the extracted first feature value and the second feature value and at least one feature value corresponding to a pre-defined voice signal and determining whether or not the audio signal of the first frame is a voice signal using the processor.

2. The method according to claim 1,

wherein the audio signal of the previous frame is a voice signal, and

the determining whether or not the audio signal of the first frame is a voice signal involves determining that the audio signal of the first frame is a voice signal when the similarity between the first feature value of the first frame and the first feature value of the previous frame is equal to or above the predetermined first threshold value.

3. The method according to claim 2,

wherein the determining whether or not the audio signal of the first frame is a voice signal comprises:

comparing a similarity between at least one of the first feature value and the second feature value and at least one feature value corresponding to the pre-defined voice signal with a predetermined second threshold value using the processor, when the similarity between the first feature value of the first frame and the first feature value of the previous frame is below the predetermined first threshold value; and

determining that the audio signal of the first frame is a noise signal when the similarity is below the predetermined second threshold value,

wherein the second threshold value is adjusted depending on whether or not the audio signal of the previous frame is a voice signal.

4. The method according to claim 1,

wherein the audio signal of the previous frame is a noise signal, and

the determining whether or not the audio signal of the first frame is a voice signal involves determining that the audio signal of the first frame is a noise signal when the similarity between the first feature value of the first frame and the first feature value of the previous frame is equal to or above the predetermined first threshold value.

5. The method according to claim 4,

wherein the determining whether or not the audio signal of the first frame is a voice signal comprises:

comparing the similarity between at least one of the first feature value and the second feature value and at least one feature value corresponding to the pre-defined voice signal with a predetermined second threshold value using the processor when the similarity between the first feature value of the first frame and the first feature value of the previous frame is below the predetermined first threshold value; and

determining that the audio signal of the first frame is a voice signal when the similarity is equal to or above the predetermined second threshold value, and

wherein the second threshold value is adjusted according to whether or not the audio signal of the previous frame is a voice signal.

6. The method according to claim 1,

wherein the determining whether or not the audio signal of the first frame is a voice signal involves, when the audio signal of the first frame is an initially input audio signal, computing a similarity between at least one of the first feature value and the second feature value of the first frame and at least one feature value corresponding to the voice signal using the processor, and comparing the computed similarity with the first threshold value using the processor, and when the similarity is equal to or above the first threshold value, determining the first frame as a voice signal.

7. The method according to claim 1,

wherein the first feature value is at least one of Mel-Frequency Cepstral Coefficients, Roll-off, and band spectrum energy.

8. The method according to claim 1,

wherein the second feature value is at least one of Low energy ratio, Zero crossing rate, Spectral flux, and Octave band energy.

9. The method according to claim 1,

wherein the determining whether or not the audio signal of the first frame is a voice signal involves, when it is determined that the audio signal of the first frame is a voice signal, classifying a speaker with respect to the audio signal of the first frame based on the extracted first feature value and the second feature value and a feature value corresponding to a pre-defined voice signal.

10. An electronic device capable of voice recognition, the device comprising:

an inputter configured to receive an input of an audio signal;

a memory configured to store at least one feature value corresponding to a pre-defined voice signal; and

a processor configured to;

when an audio signal of a first frame is input, analyze the audio signal of the first frame and extract a first feature value;

analyze the audio signal of the first frame and extract a second feature value when a similarity between the first feature value extracted from the audio signal of the first frame and a first feature value extracted from an audio signal of a previous frame is below a predetermined threshold value; and

compare the extracted first feature value and the second feature value with a feature value corresponding to a voice signal stored in the memory and determine whether or not the audio signal of the first frame is a voice signal.

11. The electronic device according to claim 10,

wherein the audio signal of the previous frame is a voice signal, and

the processor determines that the audio signal of the first frame is a voice signal when the similarity between the first feature value of the first frame and the first feature value of the previous frame is equal to or above a predetermined first threshold value.

12. The electronic device according to claim 11,

wherein, when the similarity between the first feature value of the first frame and the first feature value of the previous frame is below the predetermined first threshold value, the processor compares a similarity between at least one of the first feature value and the second feature value and at least one feature value corresponding to the pre-defined voice signal with a predetermined second threshold value, and when the similarity is below the predetermined second threshold value, the processor determines that the audio signal of the first frame is a noise signal, and

the second threshold value is adjusted depending on whether or not the audio signal of the previous frame is a voice signal.

13. The electronic device according to claim 10,

wherein the audio signal of the previous frame is a noise signal, and

the processor determines that the audio signal of the first frame is a noise signal when the similarity between the first feature value of the first frame and the first feature of the previous frame is equal to or above a predetermined first threshold value.

14. The electronic device according to claim 13,

wherein, when the similarity between the first feature value of the first frame and the first feature of the previous frame is below the predetermined first threshold value, the processor compares a similarity between at least one of the first feature value and the second feature value and at least one feature value corresponding to a pre-defined voice signal with a predetermined second threshold value, and when the similarity is equal to or above the predetermined second threshold value, determines that the audio signal of the first frame is a voice signal, and

the second threshold value is adjusted depending on whether or not the audio signal of the previous frame is a voice signal.

15. The electronic device according to claim 10,

wherein the processor, when the audio signal of the first frame is an initially input audio signal, computes a similarity between at least one of the first feature value and the second feature value of the first frame and at least one feature value corresponding to the voice signal, and compares the computed similarity with the first threshold value, and when the similarity is equal to or above the first threshold value, determines the first frame as a voice signal.

16. The electronic device according to claim 10,

wherein the first feature value is at least one of Mel-Frequency Cepstral Coefficients, Roll-off, and band spectrum energy.

17. The electronic device according to claim 10,

wherein the second feature value is at least one of Low energy ratio, Zero crossing rate, Spectral flux, and Octave band energy.

18. The electronic device according to claim 10,

wherein, when it is determined that the audio signal of the first frame is a voice signal, the processor classifies a speaker with respect to the audio signal of the first frame based on the extracted first feature value and the second feature value and a feature value corresponding to a pre-defined voice signal.

19. A non-transitory computer program combined with an electronic device and stored in a record medium in order to execute steps of:

analyzing an audio signal of a first frame when the audio signal of the first frame is input into the electronic device using an inputter of the electronic device, and extracting a first feature value using a processor of the electronic device;

determining a similarity between the first feature value extracted from the audio signal of the first frame and a first feature value extracted from an audio signal of a previous frame using the processor;

analyzing the audio signal of the first frame and extracting a second feature value when the similarity is below a predetermined threshold value using the processor; and

comparing the extracted first feature value and the second feature value and a feature value corresponding to a pre-defined voice signal, and determining whether or not the audio signal of the first frame is a voice signal using the processor.