Summary of the invention
Technical problem to be solved by this invention is: a kind of full-digital instruction hands free telephone machine is provided, adopt the digitlization noise reduction techniques of voice signal, make its signal to noise ratio that the noise suppressed threshold value can be set according to required decibel value, obtain good noise suppression effect.
The present invention addresses the above problem the technical scheme that is adopted: this full-digital instruction hands free telephone machine comprises interface circuit, codec and arithmetic and control unit, its design feature is to be provided with noise suppressor, this noise suppressor comprises serial/parallel change-over circuit, clock circuit, logical circuit and noise suppression circuit, and noise suppressor is connected with codec and arithmetic and control unit respectively.
Analog interface partly comprises two or four line change-over circuits, preamplifier, power amplifier, display interface circuit and keyboard interface circuit in the interface circuit of the present invention, preamplifier links to each other with microphone, two or four line change-over circuits link to each other with telephone wire, power amplifier links to each other with receiver, display interface circuit is connected with codec with keyboard interface circuit, and digital interface partly comprises digital interface chip DASL and preamplifier, power amplifier, display interface circuit and keyboard interface circuit.
Noise suppression circuit of the present invention mainly comprises noise suppression controller and delayer, and the noise suppression controller links to each other with delayer, noise suppression controller control noise suppressed threshold value, delay time, back delay time before the delayer control.
Noise suppressor of the present invention adopts the CPLD chip, arithmetic and control unit adopts microprocessor, codec adopts the CODEC chip, preamplifier adopts the amplifier chip, power amplifier adopts the power amplifier chip, keyboard interface circuit adopts the CPLD chip, and display interface circuit adopts the display interface special chip.
The present invention compared with prior art has the following advantages and effect: this full-digital instruction hands free telephone machine has adopted the digitlization noise reduction techniques, can carry out manually or automatically being provided with of noise suppressed threshold value according to the order of severity of the size of ambient noise, external interference and the amplitude of voice signal, thereby reach corresponding signal to noise ratio.Its interface resource is abundant, can directly substitute original dummy instruction phone, also can be used in the occasion of using new technology and new business being provided.After it carries out pcm encoder by codec CODEC chip to analog voice signal, complex programmable logic device (CPLD) with noise suppressor carries out the processing of digitlization noise suppressed to pcm stream again, when using analog interface, the PCM signal decoding is reduced to analog voice signal then; When using digital interface, the data pcm stream after the digitlization noise suppressed is handled is directly sent into the DASL interface chip.Like this, not only can obtain good noise suppression effect, but also can real-time regulated noise suppressed parameter, owing to adopt two kinds of interfaces, make things convenient for the networking and the system integration.Adopt the Audio Processing integrated circuit to carry out speech digit and handle, transmission and receive path adopt independently A/D and D/A, have overcome the voice cross-interference issue effectively, and microphone is quiet easily controls.The integrated level and the stability of this full-digital instruction hands free telephone machine all are significantly increased than prior art in addition, and cost is low.
Embodiment
The structure of embodiment is referring to Fig. 1-Fig. 3.This full-digital instruction hands free telephone machine comprises interface circuit I, codec II, noise suppressed controller III, arithmetic and control unit IV.Interface circuit I comprises two or four line change-over circuits 2,3, power amplifier 4, preamplifier 1, display interface circuit 6, keyboard interface circuit 5, digital interface circuit.Noise suppressor III comprises serial/parallel change-over circuit 7, clock circuit 8, logical circuit 9 and noise suppression circuit 10, and noise suppression circuit 10 mainly comprises noise suppression controller III and delayer 11.Preamplifier 1 links to each other with codec II with microphone, two or four line change-over circuits 2,3 link to each other with codec II with telephone wire, power amplifier 4 links to each other with codec II with receiver, display interface circuit 6 is connected with arithmetic and control unit IV with keyboard interface circuit 5, and digital interface circuit DASL links to each other with serial/parallel change-over circuit 7.Noise control inhibitor III is connected with codec II and arithmetic and control unit IV respectively.Noise suppression controller II links to each other with delayer.The important technological parameters of noise suppressed controller III control is: noise suppressed threshold value, preceding delay time, back delay time.Noise suppression controller control noise suppressed threshold value, delay time, back delay time before delayer 11 controls.
The CPLD CPLD of the noise suppressed controller III of embodiment adopts the EPM7128 chip, arithmetic and control unit IV adopts the AT90S8515 chip, CODEC codec II adopts the TP3057 chip, preamplifier 1 adopts the LM353 chip, power amplifier 4 adopts the LM3886 chip, keyboard interface circuit 5 adopts the EPM7064 chip, and display interface circuit 6 adopts the SED1335 chip, and digital interface circuit adopts TP3420.
The noise suppressed threshold value is meant the threshold level value of opening voice channel.Signal under threshold value is thought noise, closes voice channel; Signal on threshold value is then thought voice, opens voice channel, because because of noise threshold is closed with the lower part, the trend of this part is calculated to compensate by algorithm and inserted, to keep the smooth and easy sense of voice.This threshold value can manually be provided with or generate automatically according to the order of severity of the size of ambient noise, external interference and the amplitude of voice signal.For example, when the voice signal to noise ratio was 30dB, noise suppressed threshold value control point can be made as about 32mV.Because voice and two kinds of signals of noise always can not distinguish fully, therefore when signal amplitude surpasses the noise suppressed threshold value or falls back under the threshold value, delay time before need carrying out respectively and the back delay process.
Preceding delay time is meant the delay time that voice signal is opened to voice channel after surpassing threshold value.This time, the oversize initial phoneme of voice that will cause was cut, was called " head is cut ", and this is unallowable.But this time again can not be too short, and is too short, and any amplitude surpasses the of short duration interference of the burst of noise suppressed threshold value, all can open voice channel at once, and voice terminal is delivered in this interference, destroys quiet effect.Be unlikely to cause " head is cut " again for absorbing this class interference as much as possible, according to the relevant statistics and the empirical value of voice acoustic feature, preceding delay time can be selected between 0.5~4ms.In order to remedy the bottom curve that is cut, artificially increase its trend at each adopted speech waveform and extend to the bottom, calculate by its original slope, and suitably increase comfort noise that to produce undistorted effect, we are referred to as " filling a vacancy ".
Back delay time is meant at the noise suppressed thresholding and is opened and during own transmission voice, falls after rise to the noise suppressed threshold value from the voice signal amplitude, the delay time of closing to voice channel.Because the dynamic range of voice signal waveform is very big, the pause that rises and falls along with the variation of the tone again during speech, therefore back delay time too weak point can cause the interrupted of voice, influences the voice delivery quality.Back delay time is oversize, and the noise hangover influences voice quality equally when then causing speech pause.For taking into account this two aspect, the value scope of back delay time is about about 0.05~0.5s, carries out " filling a vacancy " then.
The present invention has realized the digitlization noise suppressed.Adopt the digitlization noise reduction techniques, must be earlier with digitization of speech signals.The digitlization of analog voice signal has several different methods, and the most general is carries out pcm encoder according to standard G.711.For bandwidth is the voice signal of 300~3400Hz, adopts the speed of 8kHz to carry out 8 samplings, and sampled data is by A-law encoding, and this part function is realized by CODEC codec II; 8bit serial signal from PCM decoding output DX output expression voice signal enters serial/parallel change-over circuit 7 and carries out serial/parallel conversion, and the output after the conversion is latched through eight parametrization latch modules, and every frame refreshes once.Latch signal is a frame with eight, delivers to noise suppressed controller III.Noise suppressed controller III controls voice channel, after the even bit negate with the PCM signal, again the divided-by symbol position be beyond the highest order seven bit digital with compare by the noise suppressed threshold value of setting, comparative result is exported to delayer 11, and delayer 11 outputs are then delivered to noise suppressed controller III as control signal.The PCM signal of noise suppressed controller III output, after serial/parallel change-over circuit 7 is finished parallel/serial line translation, revert to the serial pcm stream and be sent to the voice signal of CODEC decoding becoming with noise suppression effect, this partial function is finished by logic, clock, noise suppression circuit 9,8,10, by sampling, can follow the trail of ambient noise automatically and adjust output volume in addition ambient noise.Finish noise suppressed threshold value, preceding delay time, the back isoparametric setting of delay time and calculating by CPU arithmetic control circuit IV, and can extend out all digital interface circuit unit by it, for example the LCD LCD is used for doing screen demonstrations such as literal; Extend out Bus Interface Chip, can realize Long-distance Control and incision, to strengthen the additional function of full-digital instruction hands free telephone machine, i.e. extensibility by bus and far-end computer.
After adopting the digitlization noise reduction techniques, just can set and regulate the noise suppressed threshold values by software, preceding delay time, three parameters of noise suppression circuits such as back delay time, it is under different environments for use, can be by this three parameter of setting of software flexible, be that the noise suppressed threshold values can be set according to concrete applied environment, can establish little in its value of small noise condition, under the very noisy condition, reach same signal to noise ratio, its value will be established greatly, can set according to experimental data, also can set according to on-the-spot noise level, other two parameter can be set optimum value according to the speaker, to eliminate " head is cut " and noise hangover.The integrated level of equipment and stability all are significantly increased like this.
The operation principle of noise suppression circuit is referring to Fig. 2.Each part among the figure, text input Text Entry and the macroefficiency logical block Mega Function combination having adopted basic logic gate circuit, parameterized module as required respectively, represented with the default DefaultSymbol of meeting.From the 8bit serial signal of PCM decoding output DX output expression one road voice signal, the parametrization shift register module that enters 8 carries out serial/parallel conversion, and the output after the conversion is latched through 8 parametrization latch modules, and every frame refreshes once.Latch signal send with per eight and once arrives noise suppressed controller Symboll.The noise suppressed controller is controlled one road voice channel, after the even bit negate with the PCM signal, again will except that meet the position both seven bit digital the highest order with by S[6..0] the noise suppressed threshold value of setting compares, comparative result is exported to delayer 11Symbol2, and delayer 11 outputs are then delivered to noise suppressed controller III as control signal.The PCM signal of each noise suppressed controller III output reverts to the serial pcm stream and is sent to the voice signal of CODEC decoding becoming with noise suppression effect after 8 parametrization shift register modules are finished parallel/serial line translation.Symbol3 is used for producing the frame pulse F0 that CODEC requires, and shift register, latch, the required clock signal of noise suppressed controller III are provided, and provides different clock signals for delayer; TSET[3..0] be used for selecting the divide ratio of CLK1, CLK2 to adjust the noise suppressed delay time.These inputs can be provided with and regulate by external digital signal.Handle by the digitlization noise suppressed that above method is carried out the PCM signal, make voice signal produce the fixed delay of a frame 125 μ s, but the sense of hearing of people's ear can not be perceiveed fully to this time delay.
The operation principle of the delayer of noise suppression circuit is referring to Fig. 3.Wherein CLK1, CLK2 are the sprocket pulse of forward and backward time-delay, are obtained delay counter 12Countr1 and back delay counter 13Counter2 timing before being respectively applied for behind frequency division by the MCLK master clock of 2.048MHz.When no voice signal, the output D of noise suppressed controller IIISymbol1 is " 0 ", and Counter2 counts after the Q2 end is " 1 " and stops counting, and by inverter the input of CLK2 is sealed.The high level of Q2 end makes B be output as " 1 " simultaneously to the Counter1 zero clearing, and noise suppressed controller III exports PCM sign indicating number " 55H ", does not promptly have voice signal.When the PCM signal amplitude that receives surpassed the noise suppressed threshold value of setting, D became " 1 ", and Counter2 is cleared, and Counter1 disengaging this moment cleared condition begins counting.It is that " 1 " back B end output level transfers " 0 " to from " 1 " that Counter1 counts to Q1, and preceding time-delay finishes, and the output of Symbol1 becomes from " 55H " transmits the PCM sign indicating number of importing.Meanwhile, the low level of B end is sealed the CLK1 clock, as long as Counter1 is not cleared, it is low that B is always.If the PCM signal that receives is not continuous voice but burst noise, amplitude just surpasses the noise suppressed threshold value of setting momently, D becomes " 1 " back when Counter1 is not able to do in time that B become " 0 " so, D has got back to " 0 " again, the level of B is " 1 " just always, the PCM output code also is " 55H " all the time, and the noise of burst just can not pass to output.The duration of voice signal, though being lower than the noise suppressed threshold value in blink, signal amplitude make D end level be " 0 " sometimes, thereby make Counter2 break away from cleared condition sometimes and begin counting, but as long as signal amplitude is lower than the back delay time that the time of noise suppressed threshold value is no more than setting, Counter2 just always counting less than the time just by zero clearing once more, its output remains " 0 " always, make B also be " 0 " always, it is consistent that output and the PCM signal of importing remain, and voice can not continue.Have only when input signal amplitude be lower than that time of noise suppressed threshold value surpass to set after behind the delay time, Q2 export " 1 ", and Counter1 is cleared, making B is " 1 ", exports PCM sign indicating number " 55H ".When having signal again, still by above-mentioned works.