CN103761967A - Embedded speech recognition system - Google Patents
Embedded speech recognition system Download PDFInfo
- Publication number
- CN103761967A CN103761967A CN201410007573.7A CN201410007573A CN103761967A CN 103761967 A CN103761967 A CN 103761967A CN 201410007573 A CN201410007573 A CN 201410007573A CN 103761967 A CN103761967 A CN 103761967A
- Authority
- CN
- China
- Prior art keywords
- chip
- speech
- speech recognition
- recognition system
- feature
- 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
Images
Landscapes
- Power Sources (AREA)
Abstract
The invention relates to an embedded speech recognition system which comprises a master control chip, a speech processing chip, a synchronous dynamic random access memory (SDRAM) data exchange buffer chip, a speech signal collection device, a loudspeaker, a control execution module and a power module, wherein the speech signal collection device collects and transmits speech information to the speech processing chip to perform preprocessing, terminal detection and feature extraction, the speech processing chip obtains and sends feature data to the SDRAM data exchange buffer chip to achieve temporary storage and to the master control chip which obtains features and outputs a control signal after matching features with speech features in a memory feature base, the control signal is sent to the control execution module, and the power module supplies power to the chips. The control chip is applied to speech recognition to replace a personal computer (PC) in the traditional speech recognition system, therefore, the embedded speech recognition system has the advantages of small size, low cost and power consumption, flexible installation and the like, and an embedded real-time control system is utilized to manage task distribution to ensure real-time performance of recognition.
Description
Technical field
The present invention relates to a kind of speech recognition technology, particularly a kind of built-in speech recognition system.
Background technology
Built-in speech recognition system refers to that the various advanced persons' of application microprocessor realizes speech recognition technology at plate level or chip-scale software or hardware.The speech recognition system of built-in speech recognition system and PC is compared, although travelling speed and memory size have certain limitation, but it has the advantages such as volume is little, low in energy consumption, reliability is high, cost is little, flexible for installation, is specially adapted to the fields such as Smart Home, robot and consumer electronics.
Summary of the invention
The present invention be directed to built-in speech recognition system and than PC, be more suitable for the problem of household consumption, propose a kind of built-in speech recognition system, for being operated in the speech recognition system of plate level, realize the speech identifying function of efficiently and accurately.
Technical scheme of the present invention is: a kind of built-in speech recognition system, comprise main control chip, pronounciation processing chip, SDRAM exchanges data buffer chip, speech signal collection device, loudspeaker, control execution module, power module, speech signal collection device gathers transmission of speech information and carries out pre-service to pronounciation processing chip, end-point detection and feature extraction, pronounciation processing chip obtains characteristic, and to send into SDRAM exchanges data buffer chip temporary, pronounciation processing chip obtains characteristic and sends into main control chip, main control chip obtains feature and mates rear output control signal with phonetic feature in memory features storehouse, control signal is sent control execution module, power module is given each chip power supply.
Described system also comprises program download jtag interface and auxiliary circuit, and auxiliary circuit comprises the LED light of gain adjusting circuit and status recognition.
The described system integration is on a pcb board, and main control chip is connected by SPI interface with the sound identification module being comprised of pronounciation processing chip, SDRAM and peripheral auxiliary circuits.
Described pronounciation processing chip adopts MFCC algorithm to carry out feature extraction, and phonetic feature is through single order MFCC Differential Characteristics, and the single order of energy feature and energy, second order difference characteristic processing, and described main control chip adopts CHMM algorithm to carry out characteristic matching.
Beneficial effect of the present invention is: a kind of built-in speech recognition system of the present invention, control chip is applied to the PC that replaces legacy speech recognition systems in speech recognition, and make it have the advantages such as volume is little, cost is low, low in energy consumption and flexible for installation.The feature extraction algorithm of selecting and Feature Correspondence Algorithm are applicable to embedded platform of the present invention preferably, guarantee the accuracy of identification, and apply embedded real-time control system management role and distribute, and guarantee the real-time of identification.
Accompanying drawing explanation
Fig. 1 is built-in speech recognition system structural representation of the present invention;
Fig. 2 is built-in speech recognition system process flow diagram of the present invention.
Embodiment
Built-in speech recognition system structural representation as shown in Figure 1: system comprises main control chip, pronounciation processing chip, SDRAM exchanges data buffer chip, microphone, loudspeaker, control execution module, auxiliary circuit, power module, jtag interface.
Main control chip: mainly complete the phonic signal character contrast in phonic signal character to be identified and pattern base and control and carry out as core controller, can coordinate speech recognition, and the data that identification is obtained be converted to control command send control execution module complete control carry out.Wherein all programs are all take embedded real-time operating system as basic environment, realize by standard C language.
Pronounciation processing chip: as the core of identification module, mainly complete pre-service, end-point detection and feature extraction to voice signal.
Power module: for each chip provides required normal working voltage.
Microphone: speech signal collection device.
Loudspeaker: as the interactive tool of built-in speech recognition system, carry out the feedback that voice message and identification, control complete.
Auxiliary circuit: comprise gain adjusting circuit and LED light, be mainly used in the indication of gain-adjusted and status recognition.
Control execution module: after voice signal is identified, main control chip is transferred to execution module by steering order, drive control object action.
The program download interface of jtag interface: STM32.
The system integration, on a pcb board, is made it have to the advantages such as volume is little, low in energy consumption, reliability is high, cost is little, flexible for installation.Main control chip is the STM32F407VGT6 chip that ST Microelectronics produces, internal memory phonetic feature storehouse.Power circuit by input voltage voltage stabilizing to after each module required voltage value, for each module independently-powered.Main control chip is connected by SPI interface with the sound identification module being comprised of pronounciation processing chip, SDRAM and peripheral auxiliary circuits, and applies embedded real-time control system management role and distribute and control and carry out, and meets the demand of system real time.
Process flow diagram as shown in Figure 2, in dormant state, reduces power consumption when system wait triggers, and when carrying out button or password triggering, system is recovered normal operating conditions, carries out hardware and program initialization; Speech signal collection device carries out pre-service, end-point detection and feature extraction by the transmission of speech information collecting to pronounciation processing chip, will obtain characteristic and be temporarily stored in SDRAM; Main control chip will obtain feature and mate with phonetic feature in feature database, if characteristic matching is correct, will convert corresponding steering order to and will complete control action, otherwise wait for new phonetic order, no matter whether identify successfully, all give voice feedback information.
Described system adopts MFCC algorithm to carry out feature extraction, in order to improve accuracy of identification, phonetic feature has increased single order MFCC Differential Characteristics, and the single order of energy feature and energy, second order difference feature, adopt CHMM algorithm to carry out characteristic matching, with DTW, DHMM scheduling algorithm is compared, although CHMM algorithm operation quantity and memory consumption are larger, but it is applicable to unspecified person identification, arithmetic accuracy is high, identification accurately, the main control chip of native system is 32 chips simultaneously, there is flash memory on the sheet of high performance computing power and 1M byte, meet selected algorithm requirement.
Claims (4)
1. a built-in speech recognition system, it is characterized in that, comprise main control chip, pronounciation processing chip, SDRAM exchanges data buffer chip, speech signal collection device, loudspeaker, control execution module, power module, speech signal collection device gathers transmission of speech information and carries out pre-service to pronounciation processing chip, end-point detection and feature extraction, pronounciation processing chip obtains characteristic, and to send into SDRAM exchanges data buffer chip temporary, pronounciation processing chip obtains characteristic and sends into main control chip, main control chip obtains feature and mates rear output control signal with phonetic feature in memory features storehouse, control signal is sent control execution module, power module is given each chip power supply.
2. built-in speech recognition system according to claim 1, is characterized in that, described system also comprises that program downloads jtag interface and auxiliary circuit, and auxiliary circuit comprises the LED light of gain adjusting circuit and status recognition.
3. built-in speech recognition system according to claim 2, is characterized in that, the described system integration is on a pcb board, and main control chip is connected by SPI interface with the sound identification module being comprised of pronounciation processing chip, SDRAM and peripheral auxiliary circuits.
4. built-in speech recognition system according to claim 1, it is characterized in that, described pronounciation processing chip adopts MFCC algorithm to carry out feature extraction, phonetic feature is through single order MFCC Differential Characteristics, and the single order of energy feature and energy, second order difference characteristic processing, described main control chip adopts CHMM algorithm to carry out characteristic matching.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410007573.7A CN103761967A (en) | 2014-01-08 | 2014-01-08 | Embedded speech recognition system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410007573.7A CN103761967A (en) | 2014-01-08 | 2014-01-08 | Embedded speech recognition system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103761967A true CN103761967A (en) | 2014-04-30 |
Family
ID=50529195
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410007573.7A Pending CN103761967A (en) | 2014-01-08 | 2014-01-08 | Embedded speech recognition system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103761967A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105336330A (en) * | 2015-10-15 | 2016-02-17 | 上海易景信息科技有限公司 | Simple voice recognizing system |
| CN106325142A (en) * | 2015-06-30 | 2017-01-11 | 芋头科技（杭州）有限公司 | Robot system and control method thereof |
| CN107204189A (en) * | 2016-03-16 | 2017-09-26 | 中航华东光电（上海）有限公司 | The speech recognition system and method for individualized feature model can be loaded |
| CN110444204A (en) * | 2019-07-22 | 2019-11-12 | 北京艾米智能机器人科技有限公司 | A kind of offline intelligent sound control device and its control method |
| CN112804617A (en) * | 2021-01-04 | 2021-05-14 | 科大乾延科技有限公司 | Intelligent audio acquisition and processing system |
| CN112908338A (en) * | 2021-02-12 | 2021-06-04 | 深圳市众芯诺科技有限公司 | Embedded voiceprint intelligent identification chip |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102708859A (en) * | 2012-06-20 | 2012-10-03 | 太仓博天网络科技有限公司 | Real-time music voice identification system |
-
2014
- 2014-01-08 CN CN201410007573.7A patent/CN103761967A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102708859A (en) * | 2012-06-20 | 2012-10-03 | 太仓博天网络科技有限公司 | Real-time music voice identification system |
Non-Patent Citations (4)
| Title |
|---|
| 刘荣辉: "基于智能家居控制的嵌入式语音识别系统研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》, no. 10, 15 October 2013 (2013-10-15) * |
| 刘荣辉: "基于智能家居控制的嵌入式语音识别系统研究", 《全国优秀硕士学位论文全文数据库 信息科技辑》, no. 10, 15 October 2013 (2013-10-15) * |
| 杜利民: "HMM非特定人连续语音识别的嵌入式实现", 《电子与信息学报》, vol. 27, no. 1, 31 January 2005 (2005-01-31), pages 60 - 63 * |
| 陈新锐，黄理: "基于MATLAB的DHMM、DHMM和CHMM语音识别算法的对比研究", 《计算机光盘软件与应用》, no. 4, 28 February 2013 (2013-02-28), pages 68 - 69 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106325142A (en) * | 2015-06-30 | 2017-01-11 | 芋头科技（杭州）有限公司 | Robot system and control method thereof |
| CN105336330A (en) * | 2015-10-15 | 2016-02-17 | 上海易景信息科技有限公司 | Simple voice recognizing system |
| CN107204189A (en) * | 2016-03-16 | 2017-09-26 | 中航华东光电（上海）有限公司 | The speech recognition system and method for individualized feature model can be loaded |
| CN110444204A (en) * | 2019-07-22 | 2019-11-12 | 北京艾米智能机器人科技有限公司 | A kind of offline intelligent sound control device and its control method |
| CN112804617A (en) * | 2021-01-04 | 2021-05-14 | 科大乾延科技有限公司 | Intelligent audio acquisition and processing system |
| CN112908338A (en) * | 2021-02-12 | 2021-06-04 | 深圳市众芯诺科技有限公司 | Embedded voiceprint intelligent identification chip |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103761967A (en) | Embedded speech recognition system | |
| CN111708366B (en) | Robot, and method, apparatus and computer-readable storage medium for controlling movement of robot | |
| CA2973019C (en) | Control system and control method for the behavior of a robot | |
| CN111844046A (en) | Robot hardware system and robot thereof | |
| US20200234707A1 (en) | Voice interaction processing method and apparatus | |
| JP7471213B2 (en) | Voice Chips and Electronics | |
| WO2016168982A1 (en) | Method, apparatus and terminal device for setting interrupt threshold for fingerprint identification device | |
| WO2009001218A3 (en) | Electronic card able to execute a command originating from a simulation system and a command originating from a diagnostic module and associated simulation method | |
| CN105931639A (en) | Speech interaction method capable of supporting multi-hierarchy command words | |
| US11455833B2 (en) | Electronic device for tracking user activity and method of operating the same | |
| Wang et al. | Real-time block-based embedded CNN for gesture classification on an FPGA | |
| CN113678119A (en) | Electronic device for generating natural language response and method thereof | |
| CN204596410U (en) | The auditory localization of robot, wake the control device of identification up | |
| CN101620723A (en) | Offline intelligent image information processing method and intelligent image information processing system | |
| US11967322B2 (en) | Server for identifying false wakeup and method for controlling the same | |
| US20160232110A1 (en) | Input Interface Device for Portable Device | |
| CN115686198A (en) | Convenient high-precision human-computer interaction system | |
| CN104078042A (en) | Electronic device and information processing method | |
| CN207074554U (en) | A kind of more scene command word speech recognition devices | |
| CN105718019B (en) | Information processing method and electronic equipment | |
| CN105630252A (en) | Trajectory tracking camera | |
| CN201812335U (en) | Embedding type book managing device | |
| CN104102512A (en) | Embedded platform IO equipment dynamic identification system based on external interruption and IO equipment dynamic identification method of system | |
| CN202013576U (en) | Image acquisition processor of visual sense cutting robot | |
| AU2017101085A4 (en) | Control system and control method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20140430 |