CN110473544A - A kind of low-power consumption voice awakening method and device - Google Patents
A kind of low-power consumption voice awakening method and device Download PDFInfo
- Publication number
- CN110473544A CN110473544A CN201910953391.1A CN201910953391A CN110473544A CN 110473544 A CN110473544 A CN 110473544A CN 201910953391 A CN201910953391 A CN 201910953391A CN 110473544 A CN110473544 A CN 110473544A
- Authority
- CN
- China
- Prior art keywords
- voice
- low
- power consumption
- wake
- module
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000012544 monitoring process Methods 0.000 claims abstract description 5
- 238000012545 processing Methods 0.000 claims description 14
- 230000002618 waking effect Effects 0.000 claims description 6
- 239000004568 cement Substances 0.000 claims description 3
- 238000004891 communication Methods 0.000 claims description 3
- 238000013461 design Methods 0.000 claims description 3
- 238000009434 installation Methods 0.000 claims description 3
- 230000003993 interaction Effects 0.000 claims description 3
- 230000006855 networking Effects 0.000 claims description 2
- 230000005540 biological transmission Effects 0.000 claims 1
- 230000005236 sound signal Effects 0.000 description 17
- 238000001228 spectrum Methods 0.000 description 15
- 230000009467 reduction Effects 0.000 description 12
- 241000209140 Triticum Species 0.000 description 10
- 235000021307 Triticum Nutrition 0.000 description 10
- 239000007921 spray Substances 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 6
- 238000001514 detection method Methods 0.000 description 5
- 102100032219 Cathepsin D Human genes 0.000 description 4
- 101000869010 Homo sapiens Cathepsin D Proteins 0.000 description 4
- 230000009466 transformation Effects 0.000 description 4
- 230000009977 dual effect Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000009432 framing Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000037007 arousal Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000002045 lasting effect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000036651 mood Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000010183 spectrum analysis Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0208—Noise filtering
- G10L21/0216—Noise filtering characterised by the method used for estimating noise
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/78—Detection of presence or absence of voice signals
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
- G10L2015/223—Execution procedure of a spoken command
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
- G10L2015/225—Feedback of the input speech
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0208—Noise filtering
- G10L21/0216—Noise filtering characterised by the method used for estimating noise
- G10L2021/02161—Number of inputs available containing the signal or the noise to be suppressed
- G10L2021/02165—Two microphones, one receiving mainly the noise signal and the other one mainly the speech signal
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Quality & Reliability (AREA)
- Selective Calling Equipment (AREA)
Abstract
The invention discloses a kind of low-power consumption voice awakening method and device, specifically: setting acquisition module, judgment module and output module；Acquisition module and judgment module, judgment module and output module are electrically connected respectively, and acquisition module includes two highly sensitive microphones；The step of voice awakening method are as follows: in smart host off working state, real-time monitoring condition sound, when a user speaks, acquisition module acquire the voice messaging that user says, and are input to judgment module；Judgment module is compared the voice messaging of acquisition with preset wake up instruction, judges whether comprising wake up instruction；If sending wake-up signal to output module, output module sends wake-up signal to smart host, wake-up master by hardware interface comprising instruction.Hardware configuration of the present invention is simple, and smart host can be reduced power consumption, saved the energy with deep-sleep when off working state.
Description
Technical field
The present invention relates to voice technology field more particularly to a kind of low-power consumption voice awakening methods and device.
Background technique
Gradually paid attention to the continuous progress of technology by people with development, the renewable of the energy.In order to reduce to the energy
Consumption, a kind of lower standby mode of power consumption or suspend mode can be all arranged in usually many equipment, when user sets without using this
When standby, equipment can be entered in standby mode or suspend mode.Different from common operating mode, under this quasi-mode, if
Standby only to keep lower power consumption, when user needs using equipment, equipment can be carried out waking up from above-mentioned mode and be entered normally
Operating mode in.
Currently, more and more equipment are all equipped with voice Rouser or voice with the rise of voice awakening technology
Wake-up module, when user needs to wake up equipment, it is only necessary to issue corresponding voice, be connect by voice Rouser or module
The voice messaging is received, and wake operation is carried out to equipment according to voice messaging.However, in practical applications, in order to receive in time
The voice of user wakes up equipment, and voice Rouser needs to be always maintained at working condition, and power consumption is larger.In this way, just leading
Family apply during carrying out voice wake-up, the equipment actual power loss for being equipped with voice Rouser is larger.
Chinese patent 201510549435.6 provides a kind of voice awakening method and device.This method comprises: believing audio
Number carry out periodic samples, wherein obtain sampled signal in ti instance sample；Calculate the audio power of sampled signal；In audio
When energy is greater than or equal to the first threshold at ti moment, wakes up DSP and carry out voice activation detection VAD；Fail when VAD is detected,
And the difference of continuous n detection failure and the first noise energy and the first threshold at ti moment is greater than before the ti moment
When preset first threshold value, second threshold is generated according to the first noise energy, and using second threshold as the of the ti+1 moment
One threshold value, wherein the first noise energy is by being extracted with the first extraction yield 1/x to sampled signal, and to extracting
Sampled point carries out tracking filter at a slow speed and obtains.The embodiment of the present invention can reduce the number for carrying out VAD, realize terminal noisy
The reduction of power consumption under environment.
201910118663 .6 of Chinese patent, disclose a kind of directly waken up with voice is in deep sleep state
System a solution.In battery powered system or have in the application system of low-power consumption demand, does not locate in master control system
When working condition, needs master control system to be in low-power consumption standby state, reduce the stand-by power consumption electric current of system as far as possible, such as
It is that battery power supply system can greatly extend the system standby time；The system that other low-power consumption require in this way, in system standby
When reduce influence of the system standby to the other functions of system or performance as far as possible.When they need voice to wake up, need directly
System is waken up with voice signal, system is then allowed to enter normal working condition.
That there are technologies is excessively complicated for the above-mentioned prior art, manufacturing cost is high or technology is too simple, is difficult to carry out etc. asks
Topic, therefore a kind of low-power consumption voice awakening method and device easy to implement, at low cost are provided, it is to be asked with the technology of solution
Topic.
Summary of the invention
In order to solve the problems, such as to realize that above-mentioned voice arousal function power consumption is larger, the present invention provides a kind of low-power consumption voice and calls out
Awake method and device, main purpose are to reduce the actual power loss of the non-operating mode of smart machine.
In order to solve the above technical problems, the present invention provides a kind of low-power consumption voice awakening method, specifically:
Acquisition module, judgment module and output module are set；Acquisition module and judgment module, judgment module and output module difference
It is electrically connected, acquisition module includes two highly sensitive microphones；
The step of voice awakening method are as follows:
1) in smart host off working state, real-time monitoring condition sound, when a user speaks, acquisition module acquisition user are said
Voice messaging, and be input to judgment module；
2) judgment module is compared the voice messaging of acquisition with preset wake up instruction, judges whether comprising wake up instruction；
If 3) send wake-up signal to output module, output module sends wake-up signal by hardware interface comprising instruction
To smart host, wake-up master.
Further, acquisition module and judgment module use low-power consumption special digital signal processing chip.
Further, the low-power consumption special digital signal processing chip is dedicated human-computer interaction/audio processing chip,
Audio codec IP kernel, the power management module of integrated chip high-performance low-power-consumption, Codec IP be a low-power consumption, flexibly
With highly integrated stereo audio codec IP.IP supports stereo ADC and microphone or line to input, stereo DAC with
Earphone plays.
Further, the power management module, using the low power design technique based on multi-power domain and multi-clock zone,
It ensure that the low-power consumption of SoC chip.
Further, the preset wake up instruction is default wake up instruction or the customized wake up instruction of user.
Further, the specific steps of the customized wake up instruction setting of the user are as follows:
1) installation control APP is installed in smart host；
2) in smart host working condition, voice Rouser is connected to smart host, host side recalls control APP automatically；
3) it according to APP interface prompt, inputs wake up instruction and saves.
Further, after waking up smart host, voice wake up instruction receives smart host software command, continuous collecting and to sound
Sound is sent to the further identifying processing of smart host progress after carrying out the processing of the speech enhan-cements such as de-noising.
A kind of low-power consumption voice Rouser of the present invention, for the device for implementing any of the above-described low-power consumption voice awakening method.
Further, the communication of the voice Rouser and smart host is the hardware interface by smart host, hardware
Interface supports USB interface, Type-C interface, Lightning interface.
Further, the voice Rouser is powered using smart host hardware interface or internal battery, independent of
Any function or ability of the smart host of networking, work offline, and the voice messaging of acquisition does not need to upload to smart host,
To protect the privacy of user.
De-noising uses dual microphone noise reduction algorithm, specific steps are as follows:
1) 2 microphones that forward and backward placement is arranged carry out voice collecting；Preposition Mike is main Mike, is mainly responsible for voice
The detection of acquisition and spray wheat noise；Postposition Mike is auxiliary Mike, is mainly responsible for the acquisition of spray wheat noise compensation and ambient noise；
2) when voice inputs, preposition Mike and postposition Mike pickup simultaneously obtain time domain speech data T1 and T2 respectively；
3) window adding in frequency domain is carried out to the time domain speech data of preceding microphone and rear microphone respectively and Fourier transformation is handled,
Obtain frequency domain speech data F1 and F2:
4) autocorrelation spectrum PSD and coherence spectra CPSD is calculated to the frequency domain speech data of preposition Mike and postposition Mike；
5) autocorrelation spectrum PSD and coherence spectra CPSD operation relevance function are used, for judging the frequency domain of preposition Mike
The correlation of the frequency domain speech data of voice data and postposition Mike；
6) signal noise ratio SNR is estimated using relevance function, when correlation is high, the correlation height of makings and postposition Mike,
Estimate that the value of signal noise ratio is high；And when relevance function correlation is low, estimation signal noise ratio estimated value is low；And believed with estimation
It makes an uproar and calculates gain function than function；
7) gain adjustment is carried out using frequency domain speech data of the gain function to preceding microphone, the preposition Mike after obtaining noise reduction
The frequency domain speech data of wind；Frequency domain speech data after noise reduction carry out inverse Fourier transform, when frequency domain speech data are transformed into
Domain voice data；Time domain speech data after final output noise reduction；
8) analytical procedure 3) preposition Mike frequency domain speech data F1, as to belong to 20-4000hz frequency range frequency domain, amount big and for it
Even zero-decrement type, it is determined that frequency domain speech data F1 belongs to spray wheat noise；It is replaced with the frequency domain speech data F2 of postposition Mike
The spray wheat voice data for changing former preposition Mike completes the reparation that wheat noise is sprayed to preposition Mike.
Hardware configuration of the present invention is simple, and smart host can reduce power consumption with deep-sleep when off working state, saves
The energy.
Detailed description of the invention
Fig. 1 is Application Example structural schematic diagram of the present invention；
Fig. 2 is low-power consumption voice awakening method work flow diagram；
Fig. 3 is low-power consumption voice Rouser structural framing figure of the present invention.
Specific embodiment
In the following, being made a more thorough explanation with reference to attached drawing to the present invention, shown in the drawings of exemplary implementation of the invention
Example.However, the present invention can be presented as a variety of different forms, it is not construed as the exemplary implementation for being confined to describe here
Example.And these embodiments are to provide, to keep the present invention full and complete, and it will fully convey the scope of the invention to this
The those of ordinary skill in field.
The spatially relative terms such as "upper", "lower" " left side " " right side " can be used herein for ease of explanation, for saying
Relationship of the elements or features relative to another elements or features shown in bright figure.It should be understood that in addition in figure
Except the orientation shown, spatial terminology is intended to include the different direction of device in use or operation.For example, if in figure
Device is squeezed, and is stated as being located at other elements or the element of feature "lower" will be located into other elements or feature "upper".Cause
This, exemplary term "lower" may include both upper and lower orientation.Device, which can be positioned in other ways, (to be rotated by 90 ° or is located at
Other orientation), it can be interpreted accordingly used herein of the opposite explanation in space.
As shown in Figure 1 and Figure 2, a kind of low-power consumption voice awakening method of the present invention, includes the following steps；
1, in smart host off working state, acquisition module monitoring condition sound, real-time monitoring condition sound is lasting to record, and adopts
Collect environmental audio signal.When a user speaks, the voice messaging that acquisition module acquisition user says, and it is input to judgment module.
Ambient environment audio signal is acquired using dual microphone noise-reduction method, to obtain more accurately capable of speech.
In the step, while dual microphone noise reduction algorithm, specific steps are used using de-noising are as follows:
1) 2 microphones that forward and backward placement is arranged carry out voice collecting；Preposition Mike is main Mike, is mainly responsible for voice
The detection of acquisition and spray wheat noise；Postposition Mike is auxiliary Mike, is mainly responsible for the acquisition of spray wheat noise compensation and ambient noise；
2) when voice inputs, preposition Mike and postposition Mike pickup simultaneously obtain time domain speech data T1 and T2 respectively；
3) window adding in frequency domain is carried out to the time domain speech data of preceding microphone and rear microphone respectively and Fourier transformation is handled,
Obtain frequency domain speech data F1 and F2:
4) autocorrelation spectrum PSD and coherence spectra CPSD is calculated to the frequency domain speech data of preposition Mike and postposition Mike；
5) autocorrelation spectrum PSD and coherence spectra CPSD operation relevance function are used, for judging the frequency domain speech of preposition Mike
The correlation of data and the frequency domain speech data of postposition Mike；
6) signal noise ratio SNR is estimated using relevance function, when correlation is high, the correlation height of makings and postposition Mike,
Estimate that the value of signal noise ratio is high；And when relevance function correlation is low, estimation signal noise ratio estimated value is low；And believed with estimation
It makes an uproar and calculates gain function than function；
7) gain adjustment is carried out using frequency domain speech data of the gain function to preceding microphone, the preposition Mike after obtaining noise reduction
The frequency domain speech data of wind；Frequency domain speech data after noise reduction carry out inverse Fourier transform, when frequency domain speech data are transformed into
Domain voice data；Time domain speech data after final output noise reduction；
8) analytical procedure 3) preposition Mike frequency domain speech data F1, as to belong to 20-4000hz frequency range frequency domain, amount big and for it
Even zero-decrement type, it is determined that frequency domain speech data F1 belongs to spray wheat noise；It is replaced with the frequency domain speech data F2 of postposition Mike
The spray wheat voice data for changing former preposition Mike completes the reparation that wheat noise is sprayed to preposition Mike.
When it is implemented, we prepare 20000 audio files to carry out the neural network of acquisition module in the present invention
Training.The elementary audio file of 10000 voice audio files, inhuman sound frequency file as training is extracted respectively, these
20000 audio files are 2-10 seconds audio files.
10000 voice audio files include the metastable voice of audio, comprising the biggish people of the audios such as mood fluctuation
Sound.10000 inhuman sound frequency files are sound common in daily life, as field environment sound, noisy urban environment sound,
Supermarket's ambient sound, heavy rain ambient sound, bird are ambient sound, congested traffic ambient sound, construction site ambient sound etc..All sounds
The sample rate of frequency file is all 16000hz.
2, judgment module is compared the voice messaging of acquisition with preset wake up instruction, judges whether to refer to comprising waking up
It enables.
Wake up instruction is human voice signal, therefore, according to preset rules, judges whether audio signal is human voice signal.Wherein,
According to preset rules, judge whether audio signal is human voice signal, is included the following steps；
(1) the first audio signal is acquired by 2 microphones of forward and backward placement.
Preceding microphone and rear microphone obtain time domain speech data respectively, sample rate 16000hz, i.e., per second
16000 time domain speech data, we divide 128 voice data of every frame to handle, i.e., take 128 time domain speech data every time
Carry out ambient noise noise reduction.It is time domain speech data T1 and T2 that the part, which obtains output, and T1 and T2 are delayed addition respectively
Subtract each other with delay and obtain T_ADD and T_SUB, wherein T_ADD enhances for main signal, and T_SUB estimates for reference noise.It is right
Audio signal carries out spectrum analysis or extracts acoustic feature, obtains the First Eigenvalue of the first audio signal.Such as.People says
Voice signal in can decay by the degree of about 6dB/ frequency multiplication greater than the part of 800Hz, this feature value can be used as identification voice
A technical parameter.
(2) sub-frame processing is carried out to the first audio signal, obtains at least three single frames audio signal.
Framing is carried out to the first audio signal.Voice signal is divided into several segments in time-domain, per a bit of
A referred to as frame.For the voice signal in each frame, more stable signal can be regarded as.The length of usual frame voice is
10 milliseconds to 30 milliseconds.Window adding in frequency domain and Fourier transformation are carried out to the time domain speech data of the voice signal in each frame respectively
Processing obtains frequency domain speech data.Multiplying window selection is Hanning window, to every frame 128 multiplied by Hanning window coefficient, for preventing
Only below time-frequency convert when occur spectral aliasing, Fourier transformation is the conversion of time domain data to frequency domain data, and specific algorithm is real
We use Fast Fourier Transform (FFT) (FFT) in existing, to reduce the burden of hardware.
(3) the second audio signal is acquired by 2 microphones of forward and backward placement.
Second audio signal is modified the frequency domain gain of the first audio signal.Using gain function to the frequency of microphone
Domain voice data carries out gain adjustment, the frequency domain speech data of the main signal microphone after obtaining noise reduction.The input of the part is
Main signal frequency domain speech data and gain function, output are the frequency domain speech data after noise reduction.
Above-mentioned preset wake up instruction can be the instruction of default, can also be customized with user.If made by oneself using user
The preset wake up instruction specific steps of justice are as follows:
(1) installation control APP is installed in smart host；
(2) in smart host working condition, voice Rouser is connected to smart host, host side recalls control APP automatically；
(3) it according to APP interface prompt, inputs wake up instruction and saves.
If sending wake-up signal to output module, output module is sent by hardware interface and waken up 3, comprising instruction
Signal is to smart host, wake-up master.
Whether the frequency spectrum for detecting single frames audio signal frequency spectrum and human voice signal is consistent, if unanimously, audio signal is voice
Signal.
The frequency spectrum of above-mentioned single frames signal is compared with the frequency spectrum of human voice signal, whether detects single frames signal spectrum in people
Within the spectral range of acoustical signal, if audio signal comprising the single frames signal is voice within the scope of this.Such as: voice
Frequency spectrum is 6-10, when the single frames audio signal frequency spectrum of detection is 9, it is determined that corresponding voice signal is human voice signal.
In the present invention, acquisition module and judgment module use low-power consumption special digital signal processing chip.The low-power consumption is special
It is dedicated human-computer interaction/audio processing chip, the Audio of integrated chip high-performance low-power-consumption with digital signal processing chip
Codec IP kernel, power management module, Codec IP are a low-power consumption, flexible and highly integrated stereo audio encoding/decoding
Device IP.Stereo audio codec IP supports stereo ADC and microphone or line to input, and stereo DAC and earphone play.Institute
The power management module stated ensure that SoC chip using the low power design technique based on multi-power domain and multi-clock zone
Low-power consumption.
As shown in figure 3, a kind of low-power consumption voice Rouser for implementing above-mentioned low-power consumption voice awakening method, including comprising
Acquisition module 300, judgment module 310 and output module 320；Acquisition module and judgment module, judgment module and output module point
It is not electrically connected, acquisition module includes two highly sensitive microphones.Low-power consumption voice Rouser and smart host of the present invention
Communication is the hardware interface by smart host, and interface supports USB interface, Type-C interface, the types such as Lightning interface.
Further, after waking up smart host, voice wake up instruction receives smart host software command, and continuous collecting simultaneously disappears to sound
It makes an uproar after equal speech enhan-cements are handled and is sent to the further identifying processing of smart host progress.
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or spy described in conjunction with this embodiment or example
Point is included at least one embodiment or example of the invention.In the present specification, schematic expression of the above terms are not
Centainly refer to identical embodiment or example.Moreover, particular features, structures, materials, or characteristics described can be any
One or more embodiment or examples in can be combined in any suitable manner.
Although the embodiments of the present invention has been shown and described above, it is to be understood that above-described embodiment is example
Property, it is not considered as limiting the invention, those skilled in the art within the scope of the invention can be to above-mentioned
Embodiment is changed, modifies, replacement and variant.
Claims (10)
1. a kind of low-power consumption voice awakening method, which is characterized in that this method specifically:
Acquisition module, judgment module and output module are set；Acquisition module and judgment module, judgment module and output module difference
It is electrically connected, acquisition module includes two highly sensitive microphones；
The step of voice awakening method are as follows:
In smart host off working state, real-time monitoring condition sound, when a user speaks, acquisition module acquisition user are said
Voice messaging, and it is input to judgment module；
Judgment module is compared the voice messaging of acquisition with preset wake up instruction, judges whether comprising wake up instruction；
If comprising instruction, to output module send wake-up signal, output module by hardware interface transmission wake-up signal to
Smart host, wake-up master.
2. low-power consumption voice awakening method as described in claim 1, which is characterized in that acquisition module and judgment module are using low
Power consumption special digital signal processing chip.
3. low-power consumption voice awakening method as claimed in claim 2, which is characterized in that at the low-power consumption special digital signal
Chip is managed, is dedicated human-computer interaction/audio processing chip, the Audio codec IP kernel of integrated chip high-performance low-power-consumption,
Power management module, Audio codec IP are a low-power consumption, flexible and highly integrated stereo audio codec IP；
The body audio codecs IP supports stereo ADC and microphone or line to input, and stereo DAC and earphone play.
4. low-power consumption voice awakening method as claimed in claim 3, which is characterized in that the power management module uses
Low power design technique based on multi-power domain and multi-clock zone, ensure that the low-power consumption of SoC chip.
5. low-power consumption voice awakening method as described in claim 1, which is characterized in that the preset wake up instruction is silent
Recognize wake up instruction or the customized wake up instruction of user.
6. low-power consumption voice awakening method as claimed in claim 5, which is characterized in that the customized wake up instruction of user is set
The specific steps set are as follows:
Installation control APP is installed in smart host；
In smart host working condition, voice Rouser is connected to smart host, host side recalls control APP automatically；
According to APP interface prompt, inputs wake up instruction and save.
7. low-power consumption voice awakening method as described in claim 1, which is characterized in that after waking up smart host, voice is waken up
Instruction receives smart host software command, and continuous collecting simultaneously carries out sound to be sent to intelligent master after the speech enhan-cements such as de-noising are handled
Machine carries out further identifying processing.
8. a kind of low-power consumption voice Rouser, which is characterized in that called out to implement any low-power consumption voice of the claims 1-7
The device for method of waking up.
9. low-power consumption voice Rouser as claimed in claim 8, which is characterized in that the voice Rouser and intelligence are main
The communication of machine is the hardware interface by smart host, and hardware interface supports USB interface, Type-C interface, Lightning to connect
Mouthful.
10. low-power consumption voice Rouser as claimed in claim 8, which is characterized in that the voice Rouser uses intelligence
Energy host hardware interface any function or ability that perhaps internal battery is powered independent of the smart host of networking, offline
Work, the voice messaging of acquisition does not need to upload to smart host, to protect the privacy of user.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910953391.1A CN110473544A (en) | 2019-10-09 | 2019-10-09 | A kind of low-power consumption voice awakening method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910953391.1A CN110473544A (en) | 2019-10-09 | 2019-10-09 | A kind of low-power consumption voice awakening method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110473544A true CN110473544A (en) | 2019-11-19 |
Family
ID=68516930
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910953391.1A Pending CN110473544A (en) | 2019-10-09 | 2019-10-09 | A kind of low-power consumption voice awakening method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110473544A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111128164A (en) * | 2019-12-26 | 2020-05-08 | 上海风祈智能技术有限公司 | Control system for voice acquisition and recognition and implementation method thereof |
| CN111620011A (en) * | 2020-04-29 | 2020-09-04 | 苏州思必驰信息科技有限公司 | Voice processing method, intelligent container and computer readable storage medium |
| CN111986671A (en) * | 2020-08-28 | 2020-11-24 | 北京海益同展信息科技有限公司 | Service robot and voice startup and shutdown method and device thereof |
| CN112365899A (en) * | 2020-10-30 | 2021-02-12 | 北京小米松果电子有限公司 | Voice processing method, device, storage medium and terminal equipment |
| CN113990311A (en) * | 2021-10-15 | 2022-01-28 | 深圳市航顺芯片技术研发有限公司 | Voice acquisition device, controller, control method and voice acquisition control system |
| WO2023143544A1 (en) * | 2022-01-29 | 2023-08-03 | 深圳市九天睿芯科技有限公司 | Voice control method and apparatus, device, medium, and intelligent voice acquisition system |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105338459A (en) * | 2015-11-06 | 2016-02-17 | 歌尔声学股份有限公司 | MEMS (Micro-Electro-Mechanical System) microphone and signal processing method thereof |
| CN105810194A (en) * | 2016-05-11 | 2016-07-27 | 北京奇虎科技有限公司 | Voice control information acquisition method under standby state and intelligent terminal |
| CN107315519A (en) * | 2017-06-30 | 2017-11-03 | 北京奇虎科技有限公司 | OS switching methods, device and mobile terminal under driving condition |
| US20180061406A1 (en) * | 2016-08-26 | 2018-03-01 | Walton Advanced Engineering Inc. | Digitalized voice guide device and a method thereof |
| CN208316972U (en) * | 2018-07-04 | 2019-01-01 | 东莞泰升音响科技有限公司 | Has low-power consumption voice arousal function bluetooth headset |
| CN109243497A (en) * | 2018-11-02 | 2019-01-18 | 钟祥博谦信息科技有限公司 | The control method and device that voice wakes up |
| CN109741758A (en) * | 2019-01-14 | 2019-05-10 | 杭州微纳科技股份有限公司 | A kind of dual microphone voice de-noising method |
| KR20190052144A (en) * | 2016-09-29 | 2019-05-15 | 허페이 후아링 코., 엘티디. | Voice control system and its wake-up method, wake-up device and appliances, coprocessor |
| CN110012331A (en) * | 2019-04-11 | 2019-07-12 | 杭州微纳科技股份有限公司 | A kind of far field diamylose far field audio recognition method of infrared triggering |
-
2019
- 2019-10-09 CN CN201910953391.1A patent/CN110473544A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105338459A (en) * | 2015-11-06 | 2016-02-17 | 歌尔声学股份有限公司 | MEMS (Micro-Electro-Mechanical System) microphone and signal processing method thereof |
| CN105810194A (en) * | 2016-05-11 | 2016-07-27 | 北京奇虎科技有限公司 | Voice control information acquisition method under standby state and intelligent terminal |
| US20180061406A1 (en) * | 2016-08-26 | 2018-03-01 | Walton Advanced Engineering Inc. | Digitalized voice guide device and a method thereof |
| KR20190052144A (en) * | 2016-09-29 | 2019-05-15 | 허페이 후아링 코., 엘티디. | Voice control system and its wake-up method, wake-up device and appliances, coprocessor |
| CN107315519A (en) * | 2017-06-30 | 2017-11-03 | 北京奇虎科技有限公司 | OS switching methods, device and mobile terminal under driving condition |
| CN208316972U (en) * | 2018-07-04 | 2019-01-01 | 东莞泰升音响科技有限公司 | Has low-power consumption voice arousal function bluetooth headset |
| CN109243497A (en) * | 2018-11-02 | 2019-01-18 | 钟祥博谦信息科技有限公司 | The control method and device that voice wakes up |
| CN109741758A (en) * | 2019-01-14 | 2019-05-10 | 杭州微纳科技股份有限公司 | A kind of dual microphone voice de-noising method |
| CN110012331A (en) * | 2019-04-11 | 2019-07-12 | 杭州微纳科技股份有限公司 | A kind of far field diamylose far field audio recognition method of infrared triggering |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111128164A (en) * | 2019-12-26 | 2020-05-08 | 上海风祈智能技术有限公司 | Control system for voice acquisition and recognition and implementation method thereof |
| CN111128164B (en) * | 2019-12-26 | 2024-03-15 | 上海风祈智能技术有限公司 | Control system for voice acquisition and recognition and implementation method thereof |
| CN111620011A (en) * | 2020-04-29 | 2020-09-04 | 苏州思必驰信息科技有限公司 | Voice processing method, intelligent container and computer readable storage medium |
| CN111986671A (en) * | 2020-08-28 | 2020-11-24 | 北京海益同展信息科技有限公司 | Service robot and voice startup and shutdown method and device thereof |
| CN111986671B (en) * | 2020-08-28 | 2024-04-05 | 京东科技信息技术有限公司 | Service robot and voice switching-on and switching-off method and device thereof |
| CN112365899A (en) * | 2020-10-30 | 2021-02-12 | 北京小米松果电子有限公司 | Voice processing method, device, storage medium and terminal equipment |
| CN113990311A (en) * | 2021-10-15 | 2022-01-28 | 深圳市航顺芯片技术研发有限公司 | Voice acquisition device, controller, control method and voice acquisition control system |
| WO2023143544A1 (en) * | 2022-01-29 | 2023-08-03 | 深圳市九天睿芯科技有限公司 | Voice control method and apparatus, device, medium, and intelligent voice acquisition system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110473544A (en) | A kind of low-power consumption voice awakening method and device | |
| US11749262B2 (en) | Keyword detection method and related apparatus | |
| CN106714023B (en) | Bone conduction earphone-based voice awakening method and system and bone conduction earphone | |
| US9992745B2 (en) | Extraction and analysis of buffered audio data using multiple codec rates each greater than a low-power processor rate | |
| CN106782591B (en) | Device and method for improving speech recognition rate under background noise | |
| TWI474317B (en) | Signal processing apparatus and signal processing method | |
| CN105869655B (en) | Audio devices and speech detection method | |
| US9549273B2 (en) | Selective enabling of a component by a microphone circuit | |
| CN110244833B (en) | Microphone assembly | |
| US9785706B2 (en) | Acoustic sound signature detection based on sparse features | |
| US9412373B2 (en) | Adaptive environmental context sample and update for comparing speech recognition | |
| US20160071513A1 (en) | Cloud based adaptive learning for distributed sensors | |
| CN108597507A (en) | Far field phonetic function implementation method, equipment, system and storage medium | |
| JP2015501450A5 (en) | ||
| US10861462B2 (en) | Dual pipeline architecture for wakeup phrase detection with speech onset detection | |
| CN103901782A (en) | Sound control method, electronic device and sound control apparatus | |
| CN111837179A (en) | System and method for capturing noise for pattern recognition processing | |
| CN113727242B (en) | Online pickup main power unit and method and wearable device | |
| CN103778950A (en) | Recording method in the state of closed recording circuit | |
| CN103778925A (en) | Recording apparatus for controlling recording in recording circuit off state | |
| CN110191397A (en) | A kind of noise-reduction method and bluetooth headset | |
| CN111128164B (en) | Control system for voice acquisition and recognition and implementation method thereof | |
| CN104049707B (en) | Always-on low-power keyword detection | |
| WO2017118037A1 (en) | Voice-triggering method for smart device and smart device | |
| EP3125244B1 (en) | Audio classifier |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191119 |