The method of lowering noise of digital audio-frequency signal
Technical field
The present invention relates to a kind of sound signal or any method that the autocorrelation signal noise is arranged of reducing, particularly reduce sound signal or any has the noise that autocorrelation signal causes in transmission or storing process method.
Background technology
Any signal in (from the process that is sent to reception) or the stored record process (from the process that stores extraction into), always receives the erosion that medium circle body is introduced noise in transmission course, cause signal to noise ratio (S/N ratio) to reduce, and signal quality descends.As far as sound signal, signal noise increases, and signal to noise ratio (S/N ratio) reduces, and psychological auditory effect significantly descends.For example, in the application of frequency modulation type wireless microphone, because the noise in the frequency modulation wireless transmission process even microphone does not have any signal output, appoints at receiving end so to receive obvious noise (being commonly called as the singing of the stream that rustles).For another example, in traditional tape recording, because the spontaneous characteristic of tape, tape can be introduced noise, thereby has reduced the signal to noise ratio (S/N ratio) of recording signal, deterioration signal quality.
Traditional Dolby system utilizes the spectral characteristic of sound, and the higher and high-frequency energy of Low Medium Frequency energy has adopted at the preceding high frequency to sound signal of storage signal (recording) to promote with the spectral characteristic that the rising of frequency reduces.And the noise level that tape is introduced is constant; When playback, the high frequency of signal compressed and make it to return to former level, thereby the voice signal of recording is recovered reduction, the high frequency noise of introducing because of tape is simultaneously reduced by the playback circuit compression; Improve signal to noise ratio (S/N ratio), reached the effect of noise reduction.
Yet Dolby system has utilized the spectral characteristic of sound, and simultaneously, this also is its limitation.Because utilize the spectral characteristic of sound, Dolby system can't improve the signal to noise ratio (S/N ratio) of Low Medium Frequency signal.In addition, the noise reduction capability of Dolby system depends on the lifting amplitude to high-frequency signal, and the lifting amplitude is big more, noise reduction more remarkable (generally can reach 10 to 20 decibels to high-frequency signal).Yet, this to the lifting amplitude limited of high-frequency signal in the spectral characteristic and the signal level of signal: when signal level was higher, it is excessive that high-frequency signal promotes amplitude, can cause signal saturated, causes distorted signals on the contrary, therefore needs the lifting amplitude of restriction high-frequency signal; Otherwise, when signal level is low, when just signal to noise ratio (S/N ratio) is relatively poor,, the improvement of Signal-to-Noise is restricted because high-frequency signal promotes the restriction of amplitude.
Summary of the invention
For overcoming the deficiency of prior art, the object of the present invention is to provide a kind of method that reduces general simulating signal noise of causing in transmission or storing process.
In the present invention, the ultimate principle that realize to reduce the signal transmitted noise is: utilize correlation filter that signal relevant carried out analytical calculation, reduce the time correlation of signal and the quantity of information that reduces signal to reach the effect of compressed signal; Afterwards, the spread signal amplitude recently obtains best laser propagation effect with the noise that improves transmission signals as much as possible.Generally, be exactly to transmit the most simply quantity of information of (lacking), thereby improve the signal to noise ratio (S/N ratio) of transmission with maximum signal amplitude.All the other information in the original signal can be recovered through correlation filter, guarantee the fidelity of signal.
For solving the problems of the technologies described above, the present invention provides a kind of method of lowering noise of digital audio-frequency signal, it is characterized in that: comprise that delivery module transmits signal and receiver module receives two processes of signal,
Said delivery module may further comprise the steps:
1) by analog to digital converter analog signal conversion is become digital signal, deliver to transmission DSP and handle;
2) in transmitting the signal correction compression module of DSP, the coherent signal that obtains with the dependent adaptive wave filter is to the compression of being correlated with of above digital signal;
3) in the signal logarithm expansion module that transmits DSP, above signal is carried out the logarithm expansion obtain exporting digital signal;
4) in the signal log-compressed module that transmits DSP, the output digital signal is carried out log-compressed;
5) in the dependent adaptive wave filter, carry out self-adaptation correlation filtering and correction with the output digital signal of signal after the above log-compressed and delivery module, and the coherent signal that obtains is delivered to the signal correction compression module;
6) digital signal be will export by digital to analog converter and simulating signal confession emission or storage converted into;
Said receiver module may further comprise the steps:
7) simulating signal that receives or extract converts digital signal to by analog to digital converter, delivers to reception DSP and handles;
8) in the log-compressed module that receives DSP, above digital signal is carried out log-compressed;
9) with receiving original signal signal as a reference, accepting compressed signal is error signal, utilizes the sef-adapting filter principle that sef-adapting filter is carried out the self-adaptation correction, and carries out correlation filtering to receiving original signal, supplies the signal correction expansion module to use;
10) the related expanding module use accept compressed signal with correlation filtering output signal is carried out related expanding, with the signal restoring of DSP processing;
11) will reduce digital conversion of signals by digital to analog converter and become simulating signal.
The method of aforesaid lowering noise of digital audio-frequency signal is characterized in that: before the step 1) of delivery module, also comprise the step that hoisting module promotes the high frequency or the low-frequency component of simulating signal; Accordingly, after the step 11) of receiver module, comprise the step that above simulating signal is reverted to original signal through the high frequency compression filter.
The method of aforesaid lowering noise of digital audio-frequency signal is characterized in that: utilize control interface module controls log-compressed module ratio of compression.
The method of aforesaid lowering noise of digital audio-frequency signal is characterized in that: in said step 3), utilize signal energy that signal is carried out the logarithm expansion, logarithm expansion ratio is 2: 1 or 3: 1, or is any other ratio.
The method of aforesaid lowering noise of digital audio-frequency signal; It is characterized in that: in said step 5); The dependent adaptive wave filter uses the output signal of sending module to be reference signal, and the signal of log-compressed is an error signal, utilizes the sef-adapting filter principle that sef-adapting filter is carried out the self-adaptation correction; And the output signal carried out correlation filtering, supply the signal correction compression module to use.
The method of aforesaid lowering noise of digital audio-frequency signal; It is characterized in that: said correlation filter can use LS (Least Square; Least square method), LMS (Least Mean Square; Average least square method), NLMS (Normalized Least Mean Square, the average least square method of homogenization) or PNLMS adaptive algorithms such as (Proportional Normalized Least Mean Square, the average least square methods of ratio homogenization).
The method of aforesaid lowering noise of digital audio-frequency signal is characterized in that: the log-compressed ratio of signal and signal logarithm expansion ratio are complementary.
The method of aforesaid lowering noise of digital audio-frequency signal is characterized in that: in said step 6), utilize control interface module controls RF transport module signal to send duty, and emission band, frequency, power sends instruction, pilot signal etc.
The beneficial effect that the present invention reached: the method for lowering noise of digital audio-frequency signal of the present invention; Through compression and expansion to signal; Transmit the simplest quantity of information with maximum signal amplitude; Thereby improve the signal to noise ratio (S/N ratio) of transmission, all the other information in the original signal can be recovered through correlation filter simultaneously, guarantee the fidelity of signal.
Description of drawings
Fig. 1 is for transmitting the hardware structure diagram of signal processing module;
Fig. 2 is for receiving the hardware structure diagram of signal recovery module;
Fig. 3 is the schematic flow sheet of digital signal processing module in the delivery module;
Fig. 4 is the schematic flow sheet of digital signal processing module in the receiver module;
Fig. 5 is the input signal of l-G simulation test, is made up of two frequency sine waves different with amplitude, and its amplitude is respectively 16384 and 65;
Fig. 6 is equally distributed white noise for the transmitted noise of hypothesis, and its amplitude is 1000;
The transmission signal that Fig. 7 obtains after transmission digital signal processing of the present invention for the emulation input signal;
Fig. 8 is the reception signal for sending the result of signal and transmitted noise stack;
The recovering signal of Fig. 9 for obtaining after handling through receiving digital signals;
Figure 10 is Fig. 9 recovering signal expansion details of signal section significantly;
Figure 11 is the expansion details of Fig. 9 recovering signal small amplitude signal part;
Figure 12 is the expansion details of the small amplitude signal part of original signal;
Figure 13 is conventional frequency modulation transmission and the signal that obtains is formed by stacking original signal and transmitted noise;
Figure 14 is the expansion details of the signal section significantly of the conventional frequency modulation transmission signals of Figure 13;
Figure 15 is the expansion details of the small amplitude signal part of the conventional frequency modulation transmission signals of Figure 13.
Embodiment
The signal Processing of delivery module
Sound signal medium-high frequency composition is compared with the Low Medium Frequency composition; Its energy is much lower; For improving the signal to noise ratio (S/N ratio) of radio-frequency component, use the high peaking circuit that constitutes by operational amplifier that the input signal radio-frequency component is promoted, reach the purpose of the signal to noise ratio (S/N ratio) that improves total system.Usually, the audio frequency component more than the 4000Hz frequency is done 10 to 20 decibels lifting.(lifting can omitted or not have to this step.This step is corresponding with the high frequency compression filter of receiver module, does not have high boost like this step, and then the compression of the high frequency of receiver module should be 0.)
Signal after high boost obtains digital signal by analog to digital conversion, delivers to DSP and handles.Analog-to-digital sampling rate is looked the frequency band of application signal and is decided, and according to Shannon sampling rate theorem, sampling rate is at least more than 2 times of signal band.For reaching the frequency response requirement of audio frequency 20000Hz, sampling rate should adopt 44100Hz, 48000Hz, or 96000Hz etc.
According to before the output sample of signal seen off, calculate the coherent signal of prediction through the self-adaptation correlation filter.Calculate supplied with digital signal and predict the poor of coherent signal therewith, obtain the signal of relevant compression.Owing to from original signal, the relevant composition of signal is removed, reduce the time correlation of signal, thereby reduced the amplitude of signal, reached the purpose that reduces the signal message amount, played signal message amount compression effects.
In order to improve the signal to noise ratio (S/N ratio) in the signals transmission, according to the average energy of this module output signal signal is carried out the logarithm expansion, with the amplitude of the signal of expansion output as much as possible.Since the expansion of signal amplitude, and the noise level of introducing in the signals transmission is constant, thus improve the signal to noise ratio (S/N ratio) of transmission signals.The expansion of signal amplitude is a standard with the amplitude peak that system allows; Calculate ceiling capacity poor of output signal energy and system permission, the mode spread signal amplitude that this energy difference is reduced with logarithmic scale, this energy difference logarithm ratio can be 2: 1; Or 3: 1, or any other ratio or the like.For example: selecting the energy difference logarithm ratio is 2: 1 o'clock, is 10 o'clock like energy difference logarithm, the original signal amplitude is multiplied by a constant greater than 1 expands; This constant square allow the logarithm of the difference of ceiling capacity to should be 5 (10 half the) with the long-pending and system of output signal energy.This relation can be expressed from the next:
S_k=S*k;(k>1)
Signal before S representes to expand; S_k representes the signal after the logarithm expansion; K is an amplitude expansion multiple, is determined by following formula.
E_sok=E_so*k^2;
[log(E_max)-log(E_so)]∶[log(E_max)-log(E_sok)]=2∶1;
E_so representes to export the energy of signal; E_sok representes to expand the energy of back output signal; E_max representes the ceiling capacity that system allows.
For consistent, and expand complementation mutually and recovering signal, the output signal is carried out the log-compressed complementary mutually with the expansion of preceding paragraph logarithm, will export the signal compression reduction, supply the use of dependent adaptive wave filter with the preceding paragraph logarithm with the processing of receiver module.Its concrete log-compressed reduction is treated to:
S_kc=S_k/k;
S_k representes the output signal of this module of logarithm expansion back; S_kc representes the compression recovering signal after the log-compressed.The dependent adaptive wave filter uses the output signal of sending module to be reference signal; The signal of compression reduction is an error signal; Utilize the sef-adapting filter principle that sef-adapting filter is carried out the self-adaptation correction, and the output signal is carried out correlation filtering, supply the signal correction compression module to use.Sef-adapting filter can adopt LMS, NLMS, and LS, and the LMS algorithm of various variations etc.
Signal after so handling converts simulating signal into through digital to analog converter, gives the RF frequency-modulated transmitter and sees off, or be used for storage etc.Digital to analog converter need adopt the sampling rate same with analog to digital converter.Accept the signal Processing of module
Corresponding with the signal Processing of delivery module, with opposite order signal is handled, can signal be reduced.
The signal that receives is delivered to reception DSP and is handled after analog to digital conversion.This analog to digital converter should adopt the sampling rate identical with delivery module.
The reception original signal that uses the log-compressed algorithm identical with delivery module to receive compresses, and obtains accepting compressed signal.
With receiving original signal signal as a reference; Accepting compressed signal is error signal; Utilize the sef-adapting filter identical that sef-adapting filter is carried out the self-adaptation correction, and carry out correlation filtering, supply the signal correction expansion module to use receiving original signal with delivery module.
Compressed signal is accepted in the use of related expanding module and correlation filtering output is carried out related expanding to signal, with receiving the signal restoring that DSP handles.
The digital signal that receives the DSP recovery converts simulating signal into through digital to analog converter.This digital to analog converter should adopt the sampling rate identical with delivery module.
Above simulating signal reverts to original signal through the high frequency compression filter.This high frequency compression filter characteristic needs complementary mutually with the high boost filter characteristic of delivery module, makes signal restoring.Do not have high boost like delivery module, this high frequency compression can be omitted.
The control interface module is used to control RF transport module signal and sends duty, emission band, and frequency, power sends instruction, supervisory lamp; Simultaneously, log-compressed ratio in this algorithm of may command, and each parameter etc. reaches the purpose that adapts to various different application.
Below disclose the present invention with preferred embodiment, so it is not in order to restriction the present invention, and all employings are equal to replacement or the technical scheme that obtained of equivalent transformation mode, all drop within protection scope of the present invention.