CN206312566U

CN206312566U - A kind of vehicle intelligent audio devices

Info

Publication number: CN206312566U
Application number: CN201621381835.7U
Authority: CN
Inventors: 张德明
Original assignee: BEIJING SAIBIN TECHNOLOGY Co Ltd
Current assignee: BEIJING SAIBIN TECHNOLOGY Co Ltd
Priority date: 2016-12-15
Filing date: 2016-12-15
Publication date: 2017-07-07
Anticipated expiration: 2026-12-15

Abstract

The utility model proposes a kind of vehicle intelligent audio devices, including using microphone array MA, the in-car acoustical signal of pickup, echo cancellor, voice wake-up, end-point detection, Wave beam forming, second echo cancellor, ambient noise suppression, speech recognition, natural-sounding treatment are carried out to signal, corresponding actions are performed finally according to corresponding phonetic order.Using the present apparatus, the precision and recognition effect of in-car speech recognition can be obviously improved.

Description

A kind of vehicle intelligent audio devices

Technical field

The utility model belongs to electroacoustic field, more particularly to a kind of vehicle intelligent audio devices.

Background technology

With the development of vehicle intelligent system and vehicle-mounted interconnection technique, increasing automobile vendor in automobile product all Voice sound function is added, but the discrimination of existing vehicle-mounted voice device is not strong, particularly with traveling noise In the case of accuracy of identification not enough.

Utility model content

In order to overcome the defect of prior art, the utility model proposes a kind of vehicle intelligent audio devices, using the present apparatus, The precision and recognition effect of in-car speech recognition can be obviously improved.

A kind of vehicle intelligent audio devices, including microphone array MA are proposed, for picking up in-car acoustical signal, feature exists In the acoustical signal is further sent to the first echo cancellation module, for eliminating possible music background sound and communicating back Sound, voice wake-up module is entered through the acoustical signal after echo cancellor, and voice wake-up module is analyzed to the acoustical signal for entering, such as Fruit detects the acoustical signal includes default triggering voice signal, then sending instruction by voice wake-up module makes other functions mould Block reverts to working condition by low power consumpting state；After the audio devices enter into working condition, the follow-up voice letter of speaker Number endpoint detection module is further sent to, for the starting to voice signal and terminates end and detect, to reduce signal In unwanted contributions, sent to Wave beam forming module by the signal of endpoint detection module, by Wave beam forming module to microphone The signal of array pickup is processed so that the directive property of microphone array points to the person of sending of voice signal, further, voice Signal is sent to the second echo cancellation module, further to be purified to voice signal, purified voice signal send to Ambient noise cancellation module, so as to obtain pure voice signal, the voice signal is sent to sound identification module, by the mould Block is identified to voice signal, and after being analyzed further across natural-sounding processing module, obtains speaker's voice letter Command information in number, and corresponding action is performed according to the command information by voice assistant module；Intelligent audio device is also wrapped Loudspeaker SPK is included, for Play System feedback signal or machine talk to user；Action includes accordingly for wherein described execution The function of performs device itself, or interacted with vehicle, or enter row information with extraneous terminal or server using mobile network Interaction；

Wherein, microphone array includes linear array, square formation, diamond array；

In addition, sub-band filter module can be added before Wave beam forming module, carried out again after sub-band filter is carried out to signal Subsequent treatment；

Further, the device is integrated into vehicle own electronic system or is made single product, single product Including room mirror, external vehicle audio.

Brief description of the drawings

Fig. 1 shows the structural representation of vehicle intelligent audio devices of the present utility model

Specific embodiment

With reference to accompanying drawing, vehicle intelligent audio devices of the present utility model are described in detail.As shown in figure 1, this is vehicle-mounted Intelligent audio device includes microphone array MA, and for picking up in-car acoustical signal, the acoustical signal is further sent to first time Sound cancellation module (AEC) 1, for eliminating possible music background sound and communication echo, enters through the acoustical signal after echo cancellor Enter voice wake-up module (V.T.) 2, the acoustical signal of 2 pairs of entrance of voice wake-up module is analyzed, if detecting the acoustical signal Include default triggering voice signal, for example ", hello " etc., then sending instruction by voice wake-up module 2 makes other functions Module reverts to working condition by low power consumpting state；After the audio devices enter into working condition, the follow-up voice of speaker Signal is further sent to endpoint detection module (VAD) 3, is detected for the starting to voice signal and end end, with The unwanted contributions in signal are reduced, is sent to Wave beam forming module (BF) 4, by wave beam shape by the signal of endpoint detection module 3 The signal that microphone array is picked up is processed into module 4 so that the directive property of microphone array points to sending for voice signal Person, to strengthen the seizure to speaker's voice signal, further, voice signal is sent to the second echo cancellation module (AEC) 5, Further to be purified to voice signal, purified voice signal is sent to ambient noise cancellation module (ANS) 6, is used for Road is made an uproar, the steady-state noise such as wind-dryness is suppressed, so as to obtain pure voice signal, the voice signal is sent to voice knowledge Other module (ASR) 7, is identified by the module to voice signal, and is entered further across natural-sounding processing module (NLP) 8 After row analysis, the command information in speaker's voice signal is obtained, and by voice assistant module (V.A.) 9 according to the command information Corresponding action is performed, for example, some functions of itself can be performed according to command adapted thereto, or interacted with vehicle, Huo Zheli With mobile network information exchange is carried out with extraneous terminal or server；Intelligent audio device also includes loudspeaker SPK, for playing System feedback signal or machine talk are convenient for man-machine interaction to user.

The precision and identification effect of in-car speech recognition are remarkably improved using vehicle intelligent audio devices of the present utility model Really, the feature of vehicle is lifted.

Claims

1. a kind of vehicle intelligent audio devices, including microphone array MA, for picking up in-car acoustical signal, are characterised by, the sound Signal is further sent to the first echo cancellation module (1), for eliminating possible music background sound and communication echo, warp Acoustical signal after echo cancellor enters voice wake-up module (2), and voice wake-up module (2) is analyzed to the acoustical signal for entering, If detecting the acoustical signal includes default triggering voice signal, sending instruction by voice wake-up module (2) makes other Functional module reverts to working condition by low power consumpting state；After the audio devices enter into working condition, speaker is follow-up Voice signal is further sent to endpoint detection module (3), is detected for the starting to voice signal and end end, with The unwanted contributions in signal are reduced, is sent to Wave beam forming module (4), by wave beam shape by the signal of endpoint detection module (3) The signal that microphone array is picked up is processed into module (4) so that the directive property of microphone array points to the hair of voice signal The person of going out, further, voice signal is sent to the second echo cancellation module (5), further to be purified to voice signal, through carrying Pure voice signal is sent to ambient noise cancellation module (6), and so as to obtain pure voice signal, the voice signal is sent To sound identification module (7), voice signal is identified by the module, and further across natural-sounding processing module (8) After being analyzed, the command information in speaker's voice signal is obtained, and held according to the command information by voice assistant module (9) The corresponding action of row；Intelligent audio device also includes loudspeaker SPK, is extremely used for Play System feedback signal or machine talk Family.

2. vehicle intelligent audio devices as claimed in claim 1, are characterised by, described execution action accordingly includes performing The function of device itself, or interacted with vehicle, or carry out information exchange with extraneous terminal or server using mobile network.

3. vehicle intelligent audio devices as claimed in claim 1, are characterised by, microphone array includes linear array, square formation, rhombus Battle array.

4. vehicle intelligent audio devices as claimed in claim 1, are characterised by, son was added before Wave beam forming module (4) Band filtration module, subsequent treatment is carried out after sub-band filter is carried out to signal again.

5. vehicle intelligent audio devices as claimed in claim 1, are characterised by, the device is integrated into vehicle own electronic system In or be made single product.

6. vehicle intelligent audio devices as claimed in claim 4, are characterised by, single product includes room mirror, outer Put vehicle audio.