A kind of binary channels reducing noise of voice identification headset
Technical field
The utility model relates to a kind of binary channels reducing noise of voice identification headset, belongs to noise reducing apparatus field.
Background technology
Current headset noise reduction technology is divided into active noise reduction and passive noise reduction technology, and common headset all adopts wherein a kind of technology to carry out Noise reduction, meets the general object used to reach.
First the principle of passive type noise reduction headset is introduced.
Passive type noise reduction, exactly by installing sound insulation, sound-absorbing material additional, block sound wave transmission, absorb acoustic wave energy, realize human auditory system and can only receive legal transmission acoustic information, this noise reduction technology may also be referred to as physical technique noise reduction.Passive type noise reduction headset, normally by installing the outer cover of earphone additional, as cavernous body or rubber bodies etc., to reach the effect absorbing sound wave and block sound wave transmission, allows people's ear can hear the sound that the earphones such as pure music are reported.
The defect of this passive noise reduction technology is, deadens assimilation effect better for extraneous high frequency noise, but just poor for the barriering effect of low frequency noise.Most importantly, for the system gathering sound, passive noise reduction cannot be carried out.
Because these of passive noise reduction are not enough, in order to obtain better auditory effect, active noise reduction techniques arises at the historic moment.
The principle of active noise reduction headset is: utilize receiver to receive extraneous information of acoustic wave, then this information is passed to algorithm chip, the signal of telecommunication of offsetting noise information is sent after algorithm chip computing, this signal of telecommunication sends corresponding sound by a wave sounder, the sound wave that namely phase place is contrary, amplitude is identical.Bury in oblivion after this sound wave and noise sound wave meet, thus near human auditory system organ, form one noise defensive barrier, people's ear just can't hear noise.
Active noise reduction techniques may also be referred to as electronic technology noise reduction or Active Noise Control noise reduction.The common way of this technology is on headset, install one piece of Dolby circuit chip additional, and installs one piece of cell powers on headset.The scope of general noise reduction is: 100 to 1000HZ.Also have now and claim and can accomplish that noise reduction scope is at 100 to 3000HZ, but actual effect is still to be tested.The noise reduction degree of depth, generally at 12 to 18 decibels, also has and declares to accomplish 30 decibels.
Active noise reduction techniques has good effect for elimination low frequency noise.But it is just not ideal enough for elimination high frequency sound wave effect.In addition, initiatively Active Noise Control noise reduction technology has good noise reduction for the background noise of often appearance and constant appearance, for random appearance, audio frequency fluctuation is indefinite, the noise that the sound intensity is suddenly big or suddenly small, its noise reduction efficacy obviously declines, and time particularly acoustic reception is to the system of liking instead of people, the mistake caused is particularly evident.
For noise reduction headset the deficiencies in the prior art, the utility model proposes the structural measure of oneself, better noise reduction is reached by the innovation of structure, especially the acoustic information after noise reduction is provided to electronic system, desirable noise reduction can be obtained, be beneficial to the voice order that electronic system efficient identification is legal, reduce wrong identification.
Utility model content
The purpose of this utility model is to provide a kind of binary channels reducing noise of voice identification headset, to improve the accuracy rate of noise identification.
The utility model have employed following technical scheme:
A kind of binary channels reducing noise of voice identification headset, for speech recognition system, aid identification voice, it is characterized in that, comprising: support shell, nearly mouth end receiver assembly, be positioned at the lower end inside support shell, near the mouth of operator, collect the sound from operator and ambient noise; Nearly ear end receiver assembly, is positioned at the upper end outside support shell, near the ear of operator, collects the sound from operator and ambient noise; Battery, is arranged on below nearly ear end receiver assembly.
Further, binary channels reducing noise of voice identification headset of the present utility model, is characterized in that also having: compute chip assembly, is arranged in support shell.
Further, binary channels reducing noise of voice identification headset of the present utility model, it is characterized in that: wherein, compute chip assembly comprises noise filtered audio circuit, noise filtered audio circuit is connected with nearly ear end receiver assembly with nearly mouth end receiver assembly, computing identification background noise, and send contrary with background noise phase place, that amplitude the is identical anti-phase sound wave signal of telecommunication.
Further, binary channels reducing noise of voice identification headset of the present utility model, it is characterized in that: wherein, compute chip assembly comprises noise audio operational circuit, the illegal voice that computing identification receiver near the face of people is collected and burst noise, and send contrary with burst noise phase place, that amplitude the is identical anti-phase sound wave signal of telecommunication.
The beneficial effect of utility model
The purpose of this utility model aims to provide a kind of headset for speech recognition system noise reduction, that is, the service of general noise reduction headset to as if a people, namely audience is people, and the utility model service to as if a set of electronic information, be not people, audience is a set of electronic speech identifying information system.The utility model structure is simple, ambient background noise can be reduced efficiently with the illegal sound of burst to the interference of legal sound simultaneously, to overcome defect and the deficiency of above-mentioned prior art and infrastructure product, being beneficial to a set of speech recognition system can the lawful order that sends of the identification voice of precise and high efficiency, ensures industrial actual needs.
Accompanying drawing explanation
Fig. 1 is the surface structure schematic diagram of binary channels reducing noise of voice identification headset;
Fig. 2 is that binary channels headset primary clustering disassembles schematic diagram;
Fig. 3 is the structural representation of receiver assembly;
Fig. 4 is the annexation schematic diagram of other structure in compute chip and headset.
Embodiment
Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described.
As shown in Figure 1, Figure 2 and Figure 3, binary channels reducing noise of voice identification headset comprises: nearly mouth end receiver assembly 10 and nearly ear end receiver assembly 21, battery component 23, compute chip assembly 22.
Nearly mouth end receiver assembly 10, nearly mouth end receiver assembly 10, in the position near people's face, is responsible for collecting legal voice order and illegal voice order or burst noise.Legal voice order refers to the sound instruction that the people obtaining speech recognition system authentication sends; Illegal voice order refers to without speech recognition system authentication and the sound instruction that sends of the people not carrying out authorizing; Burst noise refers to that operating environment sends suddenly non-regularity, non-recurrent, signal characteristic acoustic information clearly again, the sound etc. that such as weight falls.Nearly ear end receiver assembly 21, is responsible for collecting background noise.Background noise refers in environment often exist, the sound of various sound mix, and this sound is without clear signal feature, and the sound intensity can reach 30 to 80 decibels usually.A receiver, in the outside of headset near people's ear, is responsible for collecting background noise.
As shown in Figure 4, compute chip assembly 22 adopts CSR8670.Be arranged in the headset of headset near people's ear, it comprises two major calculations circuit: background noise filtered audio circuit 42 is for computing identification background noise and send the anti-phase sound wave signal of telecommunication is contrary with background noise phase place, amplitude is identical sound wave to bury in oblivion noise.Burst noise filtered audio circuit 43, the illegal voice collected for computing identification receiver near the face of people and burst noise, and send the anti-phase sound wave signal of telecommunication is contrary with noise phase place, amplitude is identical sound wave to bury in oblivion noise, carry out filtered audio, before speech recognition system conduction, this noise is removed in legal audio signal.
Because nearly mouth end receiver assembly 10 is nearer apart from sounding position, nearly ear end receiver assembly 21 is far away apart from sounding position, therefore the instruction sound that the instruction sound nearlyer ear end receiver receipts assembly 21 that nearly mouth end receiver assembly 10 is collected is collected is strong, and due to noise source all far away apart from the distance of nearly ear end receiver assembly 21 and nearly mouth end receiver assembly 10, therefore the noise intensity that nearly mouth end receiver and nearly ear end receiver are collected is similar to identical, background noise filtered audio circuit 42 utilizes this point to distinguish background noise and instruction sound, is eliminated by background noise.
Burst noise filtered audio circuit 43, first gather one section of normal instruction sound and background noise, the audio frequency characteristics of this section of sound under extraction normal condition, then the audio frequency that collection one section is new, new audio frequency and normal audio frequency are compared, for the sound departed within the scope of normal audio, regard as burst noise and eliminate.Background noise and burst noise eliminate after sound flow to electronic system by audio output 41.
Battery component 23, is arranged in the headset of headset near people's ear, is arranged on below nearly ear end receiver assembly.This battery provides power supply for the audio operational circuit of two in compute chip assembly 22.
A kind of binary channels reducing noise of voice identification headset that the utility model provides owing to there being nearly mouth end receiver assembly 10 simultaneously, nearly ear end receiver assembly 21, compute chip assembly 22.Therefore, it is possible to better eliminate background noise and burst noise, the lawful order enabling electronic system receive voice accurately to send.