阅读说明：本技术 一种基于无线分布式麦克风阵列的语音唤醒决策系统 (Voice awakening decision-making system based on wireless distributed microphone array ) 是由 廖奎华 于 2021-03-31 设计创作，主要内容包括：本发明公开了一种基于无线分布式麦克风阵列的语音唤醒决策系统,具体涉及语音识别技术领域,包括微处理器、多个麦克风阵列和用于将麦克风阵列与控制中心连接的NB-iot模块,多个麦克风阵列采集语音信息通过NB-iot模块传输至控制中心进行处理,所述控制中心还连接有语音端点检测模块和信号处理模块,所述信号处理模块连接有声学特征提取模块,所述声学特征提取模块连接有识别解码模块。本发明能够获取更加全面的声音样本,并根据声音样本进行语音处理和分析,从而获取语音信息中的唤醒词,根据唤醒词匹配程度,完成唤醒决策内容,在常见的远场、噪音等复杂场景下,大大改善了智能语音设备的实际使用体验。(The invention discloses a voice awakening decision-making system based on a wireless distributed microphone array, and particularly relates to the technical field of voice recognition, wherein the voice awakening decision-making system comprises a microprocessor, a plurality of microphone arrays and an NB-iot module used for connecting the microphone arrays with a control center, voice information collected by the plurality of microphone arrays is transmitted to the control center for processing through the NB-iot module, the control center is further connected with a voice endpoint detection module and a signal processing module, the signal processing module is connected with an acoustic feature extraction module, and the acoustic feature extraction module is connected with a recognition decoding module. The method and the device can obtain more comprehensive sound samples, perform voice processing and analysis according to the sound samples so as to obtain the awakening words in the voice information, complete awakening decision content according to the awakening word matching degree, and greatly improve the actual use experience of the intelligent voice equipment under common complex scenes such as far fields, noises and the like.)

1. A voice awakening decision making system based on a wireless distributed microphone array is characterized in that: including microprocessor, a plurality of microphone arrays and be used for the NB-iot module of being connected microphone array and control center, a plurality of microphone arrays gather speech information and transmit to control center through NB-iot module and handle, control center still is connected with pronunciation endpoint detection module and signal processing module, signal processing module is connected with acoustic feature and draws the module, acoustic feature draws the module and is connected with the discernment decoding module, the discernment decoding module is connected with awakening word analysis module, awakening word analysis module is connected with awakening decision module.

2. The system of claim 1, wherein the system comprises: the microphone array is arranged in a voice awakening place by adopting a distributed arrangement method.

3. The system of claim 1, wherein the system comprises: the NB-iot module comprises an NB-iot uploading module, an NB-iot miniature base station and an NB-iot downloading module, and wireless distribution and data transmission of the microphone array are realized through NB-iot technology.

4. The system of claim 1, wherein the system comprises: the voice endpoint detection module carries out high-pass filtering on the received audio information, carries out voice endpoint detection according to the obtained audio data, acquires endpoint information in the voice information, and filters invalid information in the audio information according to the endpoint information.

5. The system of claim 4, wherein the system comprises: the signal processing module is also connected with the storage module, the voice information after the endpoint detection is called from the storage module, the signal processing module carries out secondary processing on the voice information, noise and invalid sound in the voice information in the two voice endpoints are filtered, and the voice information is divided into equal voice frame sections.

6. The system of claim 5, wherein the system comprises: the acoustic feature extraction module extracts features of the sound frame segments, the recognition and decoding module decodes and analyzes the extracted features to obtain awakening word information, the awakening word analysis module analyzes the awakening word information, and an awakening decision module is used for awakening a decision and awakening the system according to the analyzed similarity.

Technical Field

The invention relates to the technical field of voice recognition, in particular to a voice awakening decision system based on a wireless distributed microphone array.

Background

The voice awakening technology is an important branch in the field of voice recognition, is widely applied to voice interaction systems such as mobile phone terminals, smart homes and vehicle navigation, and is convenient for users to use voice instruction awakening equipment. More specifically, the task of the voice wake-up system is to detect a predefined wake-up word, also commonly referred to as keyword detection, from the received voice without interruption in the background, and when the system detects a corresponding keyword, the device is woken up and enters a specific working state. However, under complex scenes such as a common far field and noise, due to the complex voice, comprehensive voice data cannot be acquired and compared, so that a too high false rejection rate is caused, and the actual use experience of the intelligent voice device can be greatly reduced.

Disclosure of Invention

In order to overcome the above defects in the prior art, embodiments of the present invention provide a voice wake-up decision system based on a wireless distributed microphone array, which can obtain more comprehensive voice samples by setting distributed voice obtaining devices, perform voice processing and analysis according to the voice samples, thereby obtaining wake-up words in voice information, complete wake-up decision content according to the matching degree of the wake-up words, and greatly improve the actual use experience of intelligent voice devices in common complex scenes such as far field and noise.

In order to achieve the purpose, the invention provides the following technical scheme: the utility model provides a decision-making system is awaken to pronunciation based on wireless distributed microphone array, includes microprocessor, a plurality of microphone array and is used for the NB-iot module that is connected microphone array and control center, and a plurality of microphone array collection speech information transmit to control center through NB-iot module and handle, control center still is connected with pronunciation endpoint detection module and signal processing module, signal processing module is connected with acoustic characteristic and draws the module, acoustic characteristic draws the module and is connected with the discernment decoding module, the discernment decoding module is connected with awakening word analysis module, awakening word analysis module is connected with awakening decision-making module.

In a preferred embodiment, the microphone array is arranged at a voice wake-up site by adopting a distributed arrangement method.

In a preferred embodiment, the NB-iot modules include an NB-iot upload module, an NB-iot micro base station and an NB-iot download module, and wireless distribution and data transmission to the microphone array are realized through NB-iot technology.

In a preferred embodiment, the voice endpoint detection module performs high-pass filtering on the received audio information, performs voice endpoint detection according to the obtained audio data, acquires endpoint information in the voice information, and filters invalid information in the audio information according to the endpoint information, and is further connected with a processing module for storing and backing up the voice information after endpoint detection.

In a preferred embodiment, the signal processing module is further connected to the storage module, and retrieves the voice information after endpoint detection from the storage module, and the signal processing module performs secondary processing on the voice information, filters out noise and invalid sound in the voice information at two voice endpoints, and divides the voice information into equal amount of voice frame segments.

In a preferred embodiment, the acoustic feature extraction module extracts features of a sound frame segment, the recognition and decoding module decodes and analyzes the extracted features to obtain wakeup word information, the wakeup word analysis module analyzes the wakeup word information, and the wakeup decision module is used for performing wakeup decision to wake up the system according to the analyzed similarity.

The invention has the technical effects and advantages that:

according to the invention, more comprehensive sound samples can be obtained, and voice processing and analysis are carried out according to the sound samples, so that the awakening words in the voice information are obtained, the awakening decision content is completed according to the awakening word matching degree, the actual use experience of the intelligent voice equipment is greatly improved under the common complex scenes of far field, noise and the like, and the awakening false rejection rate is reduced.

Drawings

FIG. 1 is a schematic diagram of the system framework of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The voice wake-up decision system based on the wireless distributed microphone array shown in fig. 1 comprises a microprocessor, a plurality of microphone arrays and an NB-iot module used for connecting the microphone arrays with a control center, wherein voice information collected by the plurality of microphone arrays is transmitted to the control center through the NB-iot module for processing, the control center is further connected with a voice endpoint detection module and a signal processing module, the signal processing module is connected with an acoustic feature extraction module, the acoustic feature extraction module is connected with a recognition and decoding module, the recognition and decoding module is connected with a wake-up word analysis module, and the wake-up word analysis module is connected with a wake-up decision module;

the microphone array is arranged in a voice awakening place by adopting a distributed arrangement method;

the NB-iot module comprises an NB-iot uploading module, an NB-iot miniature base station and an NB-iot downloading module, and wireless distribution and data transmission of the microphone array are realized through NB-iot technology;

the voice endpoint detection module carries out high-pass filtering on the received audio information, carries out voice endpoint detection according to the obtained audio data, acquires endpoint information in the voice information, and filters invalid information in the audio information according to the endpoint information, and is also connected with a processing module for storing and backing up the voice information after the endpoint detection;

the signal processing module is also connected with the storage module, calls the voice information after the endpoint detection from the storage module, carries out secondary processing on the voice information, filters noise and invalid sound in the voice information in the two voice endpoints, and divides the voice information into equal voice frame sections;

the acoustic feature extraction module extracts features of the sound frame segments, the recognition and decoding module performs decoding analysis on the extracted features to acquire awakening word information, the awakening word analysis module analyzes the awakening word information, and an awakening decision module is used for carrying out awakening decision to awaken the system according to the analyzed similarity;

on the basis, when the acoustic feature extraction module performs privilege extraction on the sound frame segment, the sound frame segment is extracted based on the preset features in the system, and feature comparison is performed, so that more accurate sound features are obtained;

the preset characteristics can be trained to generate preset sound characteristics through a sound training model in the system operation process;

the recognition decoding module analyzes the acquired sound characteristics, awakening word information preset by a system is stored in the storage module, and when the recognition decoding module analyzes the sound characteristics, the characteristics are compared with the information stored in the storage module, so that the analysis is completed, and the awakening words are acquired;

the awakening decision module can match the acquired awakening words with preset awakening words, system awakening is carried out according to the matched similarity, and the system awakening is completed when the similarity is larger than 95%.

The points to be finally explained are: first, in the description of the present application, it should be noted that, unless otherwise specified and limited, the terms "mounted," "connected," and "connected" should be understood broadly, and may be a mechanical connection or an electrical connection, or a communication between two elements, and may be a direct connection, and "upper," "lower," "left," and "right" are only used to indicate a relative positional relationship, and when the absolute position of the object to be described is changed, the relative positional relationship may be changed;

secondly, the method comprises the following steps: in the drawings of the disclosed embodiments of the invention, only the structures related to the disclosed embodiments are referred to, other structures can refer to common designs, and the same embodiment and different embodiments of the invention can be combined with each other without conflict;

and finally: the above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that are within the spirit and principle of the present invention are intended to be included in the scope of the present invention.