Summary of the invention
Technology of the present invention is dealt with problems and is: overcome the deficiencies in the prior art, provide a kind of with vocal print release and the Bluetooth system that wakes up, by Application on Voiceprint Recognition, the speech recognition system in Bluetooth system is carried out release and waken up, can realize the function with voice waken system, handled easily, Application on Voiceprint Recognition also can realize identity confirmation simultaneously, thereby prevents other people usurping and false wake-up as coded lock.
The technology of the present invention solution: a kind of with vocal print release and the Bluetooth system that wakes up, comprising: audio frequency net gate way (AG) and hands-free assembly (HF) are containing two modules, i.e. AG module and HF module; HF module comprises wake module and sound identification module;
The effect of described wake module is to allow user wake speech identifying function up by phonetic order, by phonetic order, carries out bluetooth connection, realizes and calls, operates address book function; Described wake module comprises main control unit 1, for gathering voice collecting unit 2, A/D converter 3 and the vocal print recognition unit 4 of user speech; 2 output one tunnels, voice collecting unit are directly connected with an input of the main control unit 1 for the treatment of Rouser, and another road is connected with another input of main control unit 1 by A/D converter 3, Application on Voiceprint Recognition unit 4 successively; The output of main control unit 1 is connected with sound identification module input respectively, A/D converter 3 is converted to digital signal by the analoging sound signal of input, Application on Voiceprint Recognition unit mates the digital signal receiving, and matching result is sent to main control unit 1;
Described main control unit 1 receives the wake-up signal of voice collecting unit and vocal print recognition unit, can wake sound identification module up after receiving at the same time the wake-up signal of two unit, carries out speech recognition process;
Two processes are trained and identified in described Application on Voiceprint Recognition unit 4; Training process collects the voice of specifying user, and extracts the voiceprint in its voice, as password; Identifying receives speaker's voice, extracts the voiceprint in voice, using voiceprint with as the voiceprint of password, compare, if the match is successful, export wake-up signal;
Sound identification module is after system is waken up, and receives user's phonetic order, and the Word message identifying is sent to AG module by bluetooth;
AG module is carried out corresponding instruction receiving after Word message.
Described voice collecting unit 2 is comprised of the microphone for gathering user speech, audio frequency amplifier and comparator; The output of microphone is connected with the input of audio frequency amplifier, and the output of audio frequency amplifier is connected with the output of A/D converter 3 with comparator respectively; Audio frequency amplifier amplifies the audio signal of microphone output; Comparator has the threshold values of setting, this threshold values regulates according to distinct device, varying environment, when the input voltage of comparator is that the voice signal of microphone conversion is during higher than threshold values, the output of comparator output wake-up signal, Application on Voiceprint Recognition unit can be identified and identity validation the audio signal receiving simultaneously, if by authentication, also export wake-up signal, main control unit 1 is waken up after receiving the wake-up signal of comparator and voiceprint identification module simultaneously.
The present invention's advantage is compared with prior art:
(1) can realize with the voice of speech recognition Bluetooth system and waking up, not need button to wake up, full voice operation, easily.
(2) thus Application on Voiceprint Recognition can carry out authentication and avoid being usurped or false wake-up by other people.
Embodiment
As shown in Figure 1, the present invention includes: audio frequency net gate way (AG) and hands-free assembly (HF) are containing two modules, i.e. AG module and HF module; HF module comprises wake module and sound identification module;
The effect of wake module is to allow user wake speech identifying function up by phonetic order, carries out bluetooth connection, realizes call, operate address book function by phonetic order; Wake module comprises main control unit 1, for gathering voice collecting unit 2, A/D converter 3 and the vocal print recognition unit 4 of user speech; For gathering the voice collecting unit 2 of user speech, its output one tunnel is directly connected with the input of the main control unit 1 for the treatment of Rouser, another road is connected with the input of main control unit 1 by A/D converter 3, Application on Voiceprint Recognition unit 4 successively, the input of the output sound identification module 3 of main control unit 1 is connected, as shown in Figure 1.AD converter 3 is converted to digital signal by the analoging sound signal of input, and outputs to main control unit 1 by Application on Voiceprint Recognition unit 4.
As shown in Figure 1, main control unit 1 is waken up after receiving the wake-up signal of comparator and voiceprint identification module simultaneously.After waking up, can carry out operating with sound identification module.
As shown in Figure 2, wake module implementation procedure of the present invention is as follows:
(1) voice collecting unit 2 is comprised of the microphone for gathering user speech, audio frequency amplifier and comparator, the output of microphone is connected with the input of audio frequency amplifier, and the output of audio frequency amplifier is connected with the output of A/D converter 3 with comparator respectively.
(2) audio frequency amplifier amplifies the audio signal of microphone output.
(3) comparator has specific threshold values, and this threshold values can regulate according to distinct device, varying environment, when the input voltage of comparator is the voice signal of microphone conversion during higher than threshold values, and the output of comparator output wake-up signal.
(4) Application on Voiceprint Recognition unit can be identified and identity validation the audio signal receiving simultaneously, if by authentication, also exports wake-up signal.
(5) main control unit 1 is waken up after receiving the wake-up signal of comparator and vocal print recognition unit simultaneously.After waking up, can carry out operating with sound identification module.
As shown in Figure 3, Application on Voiceprint Recognition of the present invention unit implementation procedure is as follows: Application on Voiceprint Recognition needs the voiceprint that deposits nominator in advance in identification module as key, and this process is called training process.It on the process nature of identification, is the process that the voiceprint of voice that input is received and the voiceprint that prestores are compared and mated.
Training process:
(1) specify speaker to speak to voice collecting unit, the content of speaking is random, and voice collecting unit can send to voiceprint identification module by voice signal.
(2) voiceprint identification module is extracted the voiceprint of voice signal, as key, preserves.
Identifying:
(1) speaker speaks facing to voice collecting unit, and the content of speaking is random, and voice collecting unit sends to voiceprint identification module by voice signal;
(2) voiceprint identification module is extracted the voiceprint of voice signal, and compares with the vocal print key of preserving in advance;
(3) and if cipher key match, by Application on Voiceprint Recognition, output wake-up signal;
(4), if do not mated with key, do not export wake-up signal.
As shown in Figure 4, take voice, call function as example, illustrate that sound identification module of the present invention is achieved as follows:
(1) user by vocal print waken system (vocal print wakeup process describes in detail in the back) after, system can be carried out bluetooth connection to HF and AG.
(2) after connection completes, it is synchronous that system can be carried out telephone directory, by Bluetooth transmission, the telephone directory of AG end downloaded to HF end.The object of doing is like this in order to build contact identification network.
(3) build speech recognition network.Speech recognition network is divided into two parts: contact identification network and everyday words recognition network.Everyday words identification is the identification of order word, comprising: call, hang up, answer, rejection etc.Speech recognition process is the process that the characteristic information of voice is compared, mated and obtain a result with the information in recognition network of extracting.
(4), after speech recognition network struction completes, the recording of the phonetic order that user is sent is sent into speech recognition network and is identified.For example user says " phoning XXX ", and system can identify order word and " phone " in everyday words recognition network, and in contact identification network, identifies name of contact person " XXX ", thereby obtain complete recognition result, " phones XXX ".
Step (3) and step (4) complete by sound identification module.
(5) HF connects by the bluetooth of setting up the Word message identifying send to AG end, and AG end receives after Word message that transcription is corresponding command information, and carries out instruction, calls to XXX.
Can realize in the same way answering of incoming call, rejection, hangs up.
In a word, the present invention is carried out release and wakes up the speech recognition system in Bluetooth system by Application on Voiceprint Recognition, can realize the function with voice waken system, handled easily, Application on Voiceprint Recognition also can realize identity confirmation simultaneously, thereby prevents other people usurping and false wake-up as coded lock.