Method and apparatus for classifying telephone dialing test audio based on artificial intelligence

Patent metadata

Field	Value
Publication number	US-10178228-B2
Application number	US-201715789974-A
Country	US
Kind code	B2
Filing date	Oct 21, 2017
Priority date	Oct 21, 2016
Publication date	Jan 8, 2019
Grant date	Jan 8, 2019

Abstract

A method and an apparatus for classifying a telephone dialing test audio based on AI is provided. Data of a telephone dialing test audio is acquired first, then the data of the telephone dialing test audio is processed via a preset classifier so as to obtain similarities among the data of the telephone dialing test audio and telephone types in the preset classifier, in which the preset classifier is a deep-learning model determined by historical data of telephone dialing test audios and their corresponding telephone types; finally, the telephone type corresponding to the telephone dialing test audio is determined according to the similarities. With the method and apparatus, the telephone dialing test audio is classified via machine learning to determine whether a user is a normal user, thus human costs is saved and an efficiency of dialing test is increased.

First claim

What is claimed is:

1. A method for classifying a telephone dialing test audio based on artificial intelligence, comprising:
acquiring, by at least one computing device, data of a telephone dialing test audio;
processing, by the at least one computing device, the data of the telephone dialing test audio via a preset classifier, and acquiring, by the at least one computing device, similarities among the data of the telephone dialing test audio and telephone types in the preset classifier, wherein, the preset classifier is a deep-learning model determined by historical data of telephone dialing test audios and telephone types corresponding to the historical data of the telephone dialing test audios;
determining, by the at least one computing device, a telephone type corresponding to the telephone dialing test audio according to the similarities, to determine whether a user is the normal user according to the telephone dialing test audio.

2. The method according to claim 1 , wherein, before processing, by the at least one computing device, the data of the telephone dialing test audio via a preset classifier, the method further comprises:
training, by the at least one computing device, the deep-learning model via the historical data of the telephone dialing test audios and the telephone types corresponding to the historical data of the telephone dialing test audios to obtain the preset classifier.

3. The method according to claim 2 , wherein, before training, by the at least one computing device, the deep-learning model via the historical data of the telephone dialing test audios and the telephone types corresponding to the historical data of the telephone dialing test audios, the method further comprises:
extracting, by the at least one computing device, acoustic characteristics to be trained from the historical data of the telephone dialing test audios according to preset rules.

4. The method according to claim 3 , wherein, before processing, by the at least one computing device, the data of the telephone dialing test audio via a preset classifier, the method further comprises:
extracting, by the at least one computing device, effective acoustic characteristics from the data of the telephone dialing test audio according to the preset rules.

5. The method according to claim 4 , wherein, extracting, by the at least one computing device, the effective acoustic characteristics from the data of the telephone dialing test audio according to the preset rules comprises:
performing, by the at least one computing device, a voice activity detection to the data of the telephone dialing test audio so as to acquire pretreated data of the telephone dialing test audio;
cutting, by the at least one computing device, a certain length of data to be processed of the telephone dialing test audio out of the pretreated data of the telephone dialing test audio;
extracting, by the at least one computing device, the effective acoustic characteristics from the data to be processed according to a preset step size and a preset frame length.

6. The method according to claim 5 , wherein, the effective acoustic characteristics comprise N frames, the preset classifier comprises M telephone types, and acquiring, by the at least one computing device, similarities among the data of the telephone dialing test audio and telephone types in the preset classifier comprises:
putting, by the at least one computing device, the effective acoustic characteristics into the preset classifier and acquiring similarities among each of the N frames and the M telephone types;
calculating, by the at least one computing device, an average value of the similarities among the each of the N frames and the M telephone types so as to acquire M average values;
determining, by the at least one computing device, a telephone type corresponding to the telephone dialing test audio according to the similarities comprises:
determining, by the at least one computing device, a telephone type corresponding to a greatest value of the M average values as the telephone type of the telephone dialing test audio.

7. An apparatus for classifying a telephone dialing test audio based on artificial intelligence, comprising:
a processor;
a memory for storing instructions executable by the processor, wherein the processor is configured to:
acquire data of a telephone dialing test audio;
process the data of the telephone dialing test audio via a preset classifier, and acquire similarities among the data of the telephone dialing test audio and telephone types in the preset classifier, wherein, the preset classifier is a deep-learning model determined by historical data of telephone dialing test audios and telephone types corresponding to the historical data of the telephone dialing test audios;
determine a telephone type corresponding to the telephone dialing test audio according to the similarities, to determine whether a user is the normal user according to the telephone dialing test audio.

8. The apparatus according to claim 7 , wherein the processor is further configured to:
train the deep-learning model via the historical data of the telephone dialing test audios and the telephone types corresponding to the historical data of the telephone dialing test audios to obtain the preset classifier.

9. The apparatus according to claim 8 , wherein the processor is further configured to:
extract acoustic characteristics to be trained from the historical data of the telephone dialing test audios according to preset rules.

10. The apparatus according to claim 9 , wherein the processor is further configured to:
extract effective acoustic characteristics from the data of the telephone dialing test audio according to the preset rules.

11. The apparatus according to claim 10 , wherein the processor is configured to extract the effective acoustic characteristics from the data of the telephone dialing test audio according to the preset rules by acts of:
performing a voice activity detection to the data of the telephone dialing test audio so as to acquire pretreated data of the telephone dialing test audio;
cutting a certain length of data to be processed of the telephone dialing test audio out of the pretreated data of the telephone dialing test audio;
extracting the effective acoustic characteristics from the data to be processed according to a preset step size and a preset frame length.

12. The apparatus according to claim 11 , wherein, the effective acoustic characteristics comprise N frames, the preset classifier comprises M telephone types, and
the processor is configured to acquire similarities among the data of the telephone dialing test audio and telephone types in the preset classifier by acts of:
putting the effective acoustic characteristics into the preset classifier and acquiring similarities among each of the N frames and the M telephone types;
calculating an average value of the similarities among the each of the N frames and the M telephone types so as to acquire M average values;
the processor is configured to determine a telephone type corresponding to the telephone dialing test audio according to the similarities by acts of determining a telephone type corresponding to a greatest value of the M average values as the telephone type of the telephone dialing test audio.

13. A non-transitory computer readable storage medium comprising instructions, wherein when the instructions are executed by a processor of a device, the device is caused to perform a method for classifying a telephone dialing test audio based on AI, and the method comprises:
acquiring data of a telephone dialing test audio;
processing

Assignees

• Baidu online network technology beijing co ltd

Inventors

• Li Chao
• Li Xiangang
• Sun Jue

Classifications

• G06N3/045

  Combinations of networks · CPC title

• G06N3/0445

  Physics · mapped topic

• G10L25/30

  using neural networks · CPC title

• H04M3/26 Primary

  with means for applying test signals {or for measuring} · CPC title

• G10L25/78

  Detection of presence or absence of voice signals (switching of direction of transmission by voice frequency in two-way loud-speaking telephone systems H04M9/10) · CPC title