Embodiment
Below, preferred embodiment that present invention will be described in detail with reference to the accompanying.
The voice signal of the device of input among the present invention is a continuous signal (continuous signal) with the discrete signal (discrete signal) of the frequency sampling that sets in advance.
Fig. 1 is that bass strengthens the block scheme of using the overtone generating apparatus according to an embodiment of the invention.
As shown in Figure 1, the enhancing of the bass of one embodiment of the invention comprises with overtone generating apparatus 100: phasing (phasing) portion 110, fast fourier transform (FFT:fast fourier transform) portion 120, sinusoidal wave generating unit 130; Overtone generating unit 140, bandpass filtering portion 150; HRTF portion 160 and worthwhile 170.
Above-mentioned phasing portion 110 receives the multichannel voice signal by left and right sides channel, the for example input of stereophonic signal, for reducing the correlativity (correlation) between each channel sound signal, to the above-mentioned stereophonic signal of input carry out phasing and by above-mentioned left and right sides channel to above-mentioned FFT portion 120.
If stereophonic signal is carried out phasing, reduce the correlativity between the voice signal of each channel, can be by two loudspeakers to listening a taker that the sound that has more stereoscopic sensation (specious) is provided.
The above-mentioned voice signal that carries out phasing can be represented with following mathematics examination 1.
[mathematical expression 1]
Formula 1) L2=L1-aR1
Formula 2) R2=R1-aL1
At this, L1 and R1 represent the stereophonic signal of left and right sides channel input, and L2 and R2 represent the stereophonic signal of left and right sides channel output, and a represents the phasing coefficient.
Except that above-mentioned phasing method, also there is other known phasing method can realize the present invention, above-mentioned phasing method is a technique known, therefore following its explanation of omission.
Above-mentioned FFT portion 120 receives the above-mentioned voice signal that carries out phasing by above-mentioned left and right sides channel from above-mentioned phasing portion 110, voice signal to above-mentioned each channel carries out Fourier transform, for example as the high speed Fourier transform of discrete Fourier transform (DFT) (DFT), to the fourier coefficient value of the optional frequency composition of above-mentioned sinusoidal wave generating unit 130 output tut signals, above-mentioned optional frequency composition is low-frequency component preferably.
In general, music sound signal has the frequency band of OKHz to 22KHz.For aliasing (aliasing) does not take place in the above-mentioned music sound signal of sampling, with the above frequency sampling of Nyquist (Nyquist) frequency 44KHz.When utilizing when with 44KHz 1024 samples of above-mentioned music sound signal sampling being carried out FFT, the frequency interval between 1024 fourier coefficients of being calculated by above-mentioned FFT is about 40Hz.Therefore, above-mentioned low band, 7 fourier coefficients headed by becoming as the fourier coefficient of 0Hz to 300Hz audio range frequency composition.
Per 1024 samples of the voice signal of above-mentioned input are calculated 7 fourier coefficients, the fourier coefficient group of 1023 remaining voice signal samples can utilize smooth function (smoothing function) to calculate or interpolation (interplate) in i.e. one group of fourier coefficient, 1024 voice signal samples.Above-mentioned smooth function can use the linear smoothing function.For example utilize in fourier coefficient of 1024 voice signal samples one group fourier coefficient of calculating and the secondly group of utilizing next 1024 voice signal sample to calculate, the fourier coefficient of same frequency composition is connected with straight line, can obtain the fourier coefficient group of 1023 remaining voice signal samples.But,, can use many nonlinear smoothing functions of emphasizing low-frequency component for the above-mentioned fourier coefficient of interpolation certainly along with the low-frequency component of the voice signal of above-mentioned input increases.According to above-mentioned smooth function, the sinusoidal wave function that above-mentioned sinusoidal wave generating unit 130 is generated carries out Modulation and Amplitude Modulation (amplitude modulation).
Above-mentioned FFT portion 120 by 7 variant channels with above-mentioned 7 fourier coefficient values of above-mentioned each voice signal sample to sinusoidal wave generating unit 130 outputs.
Above-mentioned FFT portion 120 can adopt every 40Hz to have passband (pass band), and summit (peak) the value size of stopband (stop band) is the comb filter (multi band filter) of 13dB.Promptly, above-mentioned FFT portion 120 can comprise, the comb filter of above-mentioned execution comb filtering (not shown), with the transmission that receives 7 fourier coefficients from per 1024 the voice signal samples of above-mentioned bandpass filter, each 7 fourier coefficient of each 1023 voice signal sample that interpolation is remaining, and to the smooth function portion (not shown) of above-mentioned sinusoidal wave generating unit 130 outputs.
Above-mentioned sinusoidal wave generating unit 130 receives above-mentioned each 7 fourier coefficient value by above-mentioned each 7 channel from above-mentioned FFT portion 120, generation has 7 sine wave signals with the frequency of the corresponding size of above-mentioned each fourier coefficient value, by each 7 channel to above-mentioned overtone generating unit 140 outputs.
Can be by the sine wave signal that above-mentioned sinusoidal wave generating unit 130 generates by following mathematical expression 2 expressions.
[mathematical expression 2]
At this, k is a frequency indices, n is a time index, N is a sample number, and (k n) is the real part of k FFT coefficient generating according to smooth function to a, b (k, n) be the imaginary part of k FFT coefficient being generated by smooth function, (k is to the output of the sinusoidal wave generating unit 130 of k frequency indices in n the time index n) to lf.
Above-mentioned overtone generating unit 140 receives above-mentioned each 7 sine wave signal by above-mentioned each 7 channel from above-mentioned sinusoidal wave generating unit 130, generates the above-mentioned respectively overtone signal of 7 sine wave signals, by each independently 7 channels to above-mentioned bandpass filtering portion 150.
Above-mentioned overtone generating unit 140 is a nonlinear system, for example, has the asymmetric soft compressibility (asymmetrical soft compression system) of input-output characteristic shown in Figure 7.Above-mentioned non-linear soft compressibility is compared with other nonlinear system, generates the overtone signal of less relatively high frequency band, and concurrent looks is to less intermodulation distortion.
Above-mentioned bandpass filtering portion 150 receives the input of overtone signals from above-mentioned overtone generating unit 140 by above-mentioned each 7 channel, overtone signal to above-mentioned each channel carries out bandpass filtering, only export the multiple to set in advance in above-mentioned each channel overtone signal, the overtone signal of first and second multiple (first and secondharmonics signals) for example, 7 each channel by each separation are to 160 outputs of above-mentioned HRTF portion.
Above-mentioned HRTF portion 160 carries out the input of the overtone signal of above-mentioned bandpass filtering by above-mentioned each 7 channel, the overtone signal of above-mentioned each channel is carried out reflecting that sound handles from the HRTF that source of sound propagates into the variation of the above-mentioned overtone signal that takes place the way of listening taker, and to above-mentioned worthwhile 170 input.
If careful explanation, as shown in Figure 6, above-mentioned HRTF portion 160 is by time lag portion and subtract the portion of declining and constitute, above-mentioned time lag portion and subtract the portion's reflection that declines, by above-mentioned each channel, the time lag of the signal that the frequency content of above-mentioned overtone signal takes place during by spatial transmission and subtracting declines.Especially, above-mentioned HRTF portion 160 removes mutual sense of hearing strength difference (the inter-aural intensity difference of above-mentioned tin of taker, IID) or mutually sense of hearing time lag difference (inter-aural time difference, ITD) outside, also reflect according to the position of source of sound and different many factors.Above-mentioned characteristic is according to its characteristic difference of frequency of signal, and above-mentioned HRTF portion 160 is by above-mentioned each channel classification, has difference according to the frequency content of each overtone signal and delays coefficient N and damping coefficient S1 and S2.
Shown in Figure 6, the left side input channel of above-mentioned three-dimensional overtone signal, be that L1 is the left side audio output unit, the overtone signal that for example means left speaker output, the right side delivery channel of above-mentioned three-dimensional overtone signal, be that R1 is the right side audio output unit, for example mean the overtone signal of right speaker output.Above-mentioned S1 represents that signal declines from the subtracting of signal that the rectilinear propagation of tut efferent takes place when listening taker, and above-mentioned S2 represents that the subtracting of signal that signal takes place declines when tut efferent diagonal line propagates into above-mentioned tin of taker.That is, above-mentioned S1 and S2 reflect above-mentioned IID.In addition, the above-mentioned coefficient N reflection ITD that delays, the time lag that takes place when promptly propagating for the diagonal line of listening taker to compare with signal rectilinear propagation.
Average gain (average gain) to the HRTF of the frequency band under above-mentioned each overtone signal is set to above-mentioned S1 and S2, and the delay time of the above-mentioned HRTF of frequency band under above-mentioned each overtone signal is set to above-mentioned N.
Above-mentioned worthwhile 170 worthwhile above-mentioned each channel overtone signal and the output of handling through HRTF from 160 inputs of above-mentioned HRTF portion.
In general, overtone is that signal with low-pass filtering is input to nonlinear system and generates.The inapplicable superposition theorem of nonlinear system (law of superposition) if the signal that input has many frequency contents generates the intermodulation composition, and then does not have in the signal that generates and adds at expected frequency signal, and intermodulation distortion takes place.For example, when nonlinear system half-wave rectifier (half-wave rectifier) input comprised the signal of DC composition, the gross energy of the overtone of generation depended on the energy rank of DC, and worst does not generate any overtone.
For minimizing the generation of above-mentioned intermodulation composition, the signal segmentation that is input to above-mentioned nonlinear system become narrow band signal after, the signal of respectively cutting apart is input to nonlinear system.For this reason, can use comb filter, but this comb filter has higher complexity with meticulousr frequency characteristic, and difficult the realization.
Above-mentioned bass strengthens the comb filter that the above-mentioned sinusoidal wave generating unit 130 that adopts the above-mentioned FFT120 that carries out Fourier transforms and utilize fourier coefficient to generate sine wave signal with overtone generating apparatus 100 replaces having above-mentioned meticulous frequency characteristic, provides to have the also overtone generating apparatus of easier realization of relatively low complexity.
Above-mentioned bass enhancing can be used for generating the overtone signal of radio-frequency component with overtone generating apparatus 100, but preferably is used to generate the overtone signal of low-frequency component.Because the low-frequency band that is used for the bass enhancing has the 300Hz size of 0Hz to 300Hz, the high frequency band that is used for the high pitch enhancing on the contrary has the 10KHz size of 12KHz to 22KHz.For the wave band with 10KHz size is all carried out enhancement process,, need to adopt, and need to use different channels by each fourier coefficient than above-mentioned 7 more fourier coefficients of fourier coefficient if adopt above-mentioned enhancing with overtone device 100.Therefore, if will realize above-mentioned overtone generating apparatus 100, need to use device with quite high calculation process ability in order to strengthen high pitch.
Fig. 2 is that bass strengthens the block scheme of using the overtone generating apparatus according to another embodiment of the present invention.
Strengthen with overtone generating apparatus 100 according to the bass of this embodiment and to strengthen with overtone generating apparatus 100 with the bass of Fig. 1 substantially, therefore can attach most importance to distinctive points describes, and identical Reference numeral is represented identical structure.
As shown in Figure 2, the bass of another embodiment of the present invention strengthens and to comprise with overtone generating apparatus 100, phasing portion 110, FFT portion 120, sinusoidal wave generating unit 130, set portion 135, overtone generating unit 140, bandpass filtering portion 150, HRTF portion 160 and worthwhile 170.
FFT portion 120 as shown in Figure 1, above-mentioned FFT portion 120 receives the voice signal of above-mentioned phasing from above-mentioned phasing portion 110 by above-mentioned left and right sides channel, voice signal to each channel carries out FFT, after calculating 7 fourier coefficient values to the low-frequency band composition of tut signal, by each independently 7 channels with above-mentioned 7 fourier coefficient values to above-mentioned sinusoidal wave generating unit 130 outputs.
Above-mentioned sinusoidal wave generating unit 130 by each independently 7 channels sine wave signal that will have corresponding size of above-mentioned each fourier coefficient value and a frequency export to above-mentioned set portion 135.
Above-mentioned set portion 135 receives the sine wave signal of above-mentioned generation from above-mentioned sinusoidal wave generating unit 130 by above-mentioned each channel, the sine wave signal of a part of channel in each sine wave signal is synthesized a signal, to above-mentioned overtone generating unit 140 outputs, the signal of all the other channels still passes through each channel to above-mentioned overtone generating unit 140 by a channel.
In above-mentioned set portion 135, the reason that will synthesize a channel output by the sine wave signal of plural channel output is to reduce the quantity of above-mentioned channel, and reduce the calculation amount of above-mentioned overtone generating apparatus 100.But, when the number that reduces above-mentioned channel and when being input to above-mentioned overtone generating apparatus 100, comprise more intermodulation composition in the overtone signal that generates by above-mentioned synthetic channel, therefore bigger intermodulation distortion phenomenon takes place.
During people's perceived sounds signal, be divided into special frequency band classification and perception, in above-mentioned special frequency band, even there is the intermodulation composition that generates according to the overtone signal, can't sharp perceiving.Therefore, when according to the synthetic above-mentioned channel of above-mentioned special frequency band and when generating above-mentioned overtone signal, the bass that can realize that the calculation amount is less relatively and listen taker can't discern the difference of intermodulation distortion strengthens with overtone generating apparatus 100.
Fig. 3 is that bass strengthens the block scheme of using the overtone generating apparatus according to another embodiment of the present invention.
Bass enhancing according to present embodiment is identical with overtone generating apparatus 100 with the bass enhancing of Fig. 1 substantially with overtone generating apparatus 100, therefore can stress distinctive points, and same reference numerals is represented identical structure.
As shown in Figure 3, the bass of another embodiment of the present invention strengthens and to comprise with the overtone generating apparatus, phasing portion 110, FFT portion, low-pass filter 125, sinusoidal wave generating unit 130, overtone generating unit 140, bandpass filtering portion 150, HRTF portion 160 and worthwhile 170.
Above-mentioned low-pass filter 125 is by the signal of above-mentioned left and right sides channel from the above-mentioned phasing of above-mentioned phasing portion's 110 receptions, voice signal to above-mentioned each channel carries out low-pass filtering, only export the ultra-low frequency signal (sub-bass frequency signal) that sets in advance, export to above-mentioned overtone generating unit 140 by above-mentioned each left and right sides channel.The preferably unlimited pulsion phase of above-mentioned low-pass filter 125 should (IIR:infinite impulse response) wave filter.
FTT portion 120 as shown in Figure 1, above-mentioned FFT portion 120 receives the voice signal of above-mentioned phasing by left and right sides channel from above-mentioned phasing portion 110, and the voice signal of above-mentioned each channel is carried out FFT, calculates 7 fourier coefficient values to the low-frequency component of tut signal.Afterwards, above-mentioned FFT portion 120 by each independently the channel fourier coefficient that will not belong to the medium and low frequency composition (mid-bass frequency components) of above-mentioned ultralow frequency band output to above-mentioned sinusoidal wave generating unit 130.
Above-mentioned sinusoidal wave generating unit 130 by above-mentioned each independently channel receive the fourier coefficient values from above-mentioned FFT portion 120, generation has the sine wave signal of above-mentioned corresponding size of each fourier coefficient value and frequency, by each independently channel output to above-mentioned overtone generating unit 140.
Above-mentioned overtone generating unit 140 receives from above-mentioned low-pass filter 125 by a channel above-mentionedly carries out the voice signal of low-pass filtering by left and right sides channel, and generate above-mentioned input voice signal frequency content the overtone signal and output to above-mentioned bandpass filtering portion 150 by a channel.In addition, above-mentioned overtone generating unit 140 receives above-mentioned sine wave signal by above-mentioned each channel from above-mentioned sinusoidal wave generating unit 130, generates the overtone signal to the sine wave signal of above-mentioned each channel, by each independently channel output to above-mentioned bandpass filtering portion 150.
Above-mentioned bass strengthens with overtone generating apparatus 100 different with overtone generating apparatus 100 as shown in Figure 1, the frequency content that belongs to ultralow frequency band is not carried out Fourier transform, the independent overtone signal that generates, be input to above-mentioned overtone generating unit 140 and generate the overtone signal by a channel.On the contrary,,, the sine wave signal that utilizes Fourier transform to generate is input to above-mentioned overtone generating unit 140, generates the overtone signal as overtone generating apparatus 100 shown in Figure 1 to belonging to the frequency content of medium and low frequency band.
It has been generally acknowledged that the frequency content of ultralow frequency band is sensation rather than listens the mankind's ear.That is, when voice signal belongs to ultralow frequency band, listen taker as the frequency content that belongs to the medium and low frequency band, to perceive observantly.Therefore, can realize that a kind of bass strengthens with overtone generating apparatus 100, after it carries out low-pass filtering and obtain the frequency content of ultralow frequency band voice signal, generates the overtone signal by a channel, therefore the calculation amount is less relatively, listens taker can't feel the difference of intermodulation distortion.
Fig. 4 is that high pitch strengthens the block scheme of using the overtone generating apparatus according to an embodiment of the invention.
As shown in Figure 4, the enhancing of the high pitch of one embodiment of the invention comprises with overtone generating apparatus 200: phasing portion 210, the first bandpass filter 220, and reduce sampling and commands troops 230, frame repetition portion 240, partes glabra 250, the second bands are with wave filter 260 and HRTF portion 270.
Above-mentioned phasing portion 210 receives the multichannel voice signal by left and right sides channel, for example, the input of stereophonic signal is for reducing each channel sound correlation between signals, above-mentioned stereophonic signal to input carries out phasing, outputs to above-mentioned first bandpass filter 220 by above-mentioned left and right sides channel.Therefore above-mentioned phasing portion's 210 execution and Fig. 1, below omit its detailed description to phasing portion 110 identical functions and action shown in Figure 3.
Above-mentioned first bandpass filter 220 is from the above-mentioned voice signal that carries out phasing of above-mentioned phasing portion's 210 receptions, the tut signal is carried out bandpass filtering, only export the passband that sets in advance, for example the frequency content of 6KHz to 11KHz wave band outputs to above-mentioned reduction sampling and commands troops 230.The minimum of the passband of above-mentioned low 1 bandpass filter 220 and highest frequency preferably with the minimum of the frequency band under the overtone signal of desire output and highest frequency value divided by the command troops value of 230 reduction sampling rate multiple of above-mentioned reduction sampling, its reasons are as follows state bright.
The above-mentioned reduction sampling 230 pairs of voice signals from 220 inputs of above-mentioned first bandpass filter of commanding troops reduce sampling rate with the frame sign that sets in advance, the multiple that sets in advance, for example adopt 1024 samples or adopt 2 times, generate and output to above-mentioned frame repetition portion 240 behind the overtone signal of voice signal of above-mentioned input.
Command troops 230 input audio signal of above-mentioned reduction sampling is X (z), and when the voice signal that reduces sampling rate was Y (z), the voice signal of above-mentioned reduction sampling rate was according to following mathematical expression 3 definition.
[mathematical expression 3]
Wherein, X (z) is an input signal, and Y (z) is the voice signal that reduces sampling rate, and z is a z zone parameter, and n is a number of samples, and M is a decline sampling rate multiple.
That is, above-mentioned reduction sampling commands troops 230 in per 2048 samples of the voice signal of above-mentioned input, and only first 1024 samples of sampling are abandoned all the other 1024 samples.The voice signal of above-mentioned reduction sampling rate comprises the overtone signal, and this overtone signal has frequency content high above-mentioned reduction sampling rate multiple, for example 2 times the frequency than the voice signal that is included in above-mentioned input.
Therefore, in the time will strengthening the overtone signal of using overtone generating apparatus 200 generation 12KHz to 22KHz by above-mentioned high pitch, the band that above-mentioned first bandpass filter 220 preferably has 6KHz to 11KHz is logical.
When voice signal being reduced sampling rate and generating the overtone signal, the situation that generates the overtone signal with other nonlinear system of sampling is different, can generate the preferable overtone signal that does not have the intermodulation composition, do not have intermodulation distortion.
Above-mentioned frame repetition portion 240 carries out the frame re-treatment from the above-mentioned reduction sampling 230 overtone signals that receive above-mentioned reduction sampling of commanding troops to the overtone signal of above-mentioned generation, outputs to above-mentioned partes glabra 250 behind the above-mentioned overtone signal of interpolation.
That is, blank parts as the frame sign that takes place owing to above-mentioned reduction sampling rate is filled up with the voice signal sample of frame before the sampling by above-mentioned frame repetition portion 240 in the overtone signal of above-mentioned input.For example, in per 2048 samples of the voice signal of above-mentioned input, fill up by the above-mentioned reduction sampling 230 residue blank parts of abandoning of commanding troops with 1024 samples of head of sampling.
Above-mentioned partes glabra 250 receives the input of the voice signal that strengthens from above-mentioned frame repetition portion 240, the overtone signal of above-mentioned interpolation is carried out smoothing processing and output, this smoothing processing is exactly the part that frame repeats in the level and smooth above-mentioned overtone signal, i.e. frame boundaries (frame edge boundary).
Its purpose is to prevent the rapid variation of the part generation signal that repeats at above-mentioned frame, exports more natural voice signal.The mean value of the last sample of above-mentioned partes glabra 250 each frame and first and second sample of next frame be set to next frame the first time sample the voice signal value.
When carrying out above-mentioned frame re-treatment, because frame boundaries can generate the unwanted frequency composition.The frame that the size of use frame and the cycle of first-harmonic are bigger is also carried out above-mentioned smoothing processing, can prevent above-mentioned frame boundaries effect.
Above-mentioned second bandpass filter 260 receives the overtone signal of above-mentioned interpolation from above-mentioned partes glabra 250, signal to above-mentioned interpolation is carried out bandpass filtering, only export the overtone signal of the multiple that sets in advance in the overtone signal of above-mentioned interpolation, the first overtone signal for example outputs to above-mentioned HRTF portion 270 then.Its purpose is the frequency content except that removing needed overtone signal.
When strengthening when generating the overtone signal of 12KHz to 22KHz for example with overtone generating apparatus 200 by above-mentioned high pitch, above-mentioned second bandpass filter 220 preferably has the passband of 12KHz to 22KHz.
Above-mentioned HRTF portion 270 receives above-mentioned overtone signal from above-mentioned second bandpass filter 260, the overtone signal of above-mentioned input carried out HRTF handles and output, this HRTF handles reflection sound and propagates into the variation of the above-mentioned overtone signal that takes place the way of listening taker from source of sound.
Above-mentioned HRTF portion 270 is by time lag portion and subtract the portion of declining and constitute, above-mentioned time lag portion and subtract the portion's reflection that declines, and by above-mentioned each channel, the time lag of the signal that the frequency content of above-mentioned overtone signal takes place during by spatial transmission and subtracting declines.Above-mentioned HRTF portion 270 carries out with Fig. 1 and carries out identical functions and actions to phasing portion shown in Figure 3 110, therefore followingly omits its detailed description.
Usually, when restoring according to the music file of compression algorithms such as MP3 compression and playing, can lose and belong to high frequency band, for example the about frequency content of the above high frequency band of 12KHz or 15KHz can't correctly be play former sound.The information of frequency content that belongs to the high frequency band of above-mentioned forfeiture does not exist, and therefore has no idea to play former sound self.But quite similarly possibility is very high to utilize the frequency content signal of frequency content overtone signal that generates and the high frequency band that belongs to above-mentioned forfeiture of the voice signal of above-mentioned recovery.Therefore, when using above-mentioned high pitch to strengthen, can play the voice signal of comparing more faithful to former sound with high frequency band with overtone generating apparatus 200 interpolation voice signals.
Fig. 8 is that high pitch strengthens the voice signal spectrum analysis figure with the input and output of overtone generating apparatus 200.At first, the longitudinal axis is represented size, and transverse axis is represented frequency.As shown in Figure 8, the signal that will have the frequency content of 8KHz and 9KHz is input to 2 times of above-mentioned high pitchs that reduce sampling rate and strengthens with overtone generating apparatus 200, and the result can learn the overtone signal of having exported the frequency content with 16KHz and 18KHz.The size of other intermodulation compositions is very little, can ignore.If use other nonlinear system to replace above-mentioned high pitch to strengthen, the quite intermodulation composition of size of 17KHz can occur with overtone generating apparatus 200.
Above-mentioned overtone generating apparatus 200 more is applicable to the frequency content that strengthens high frequency band.Because during low band signal, the quantity of information that comprises in identical time interval is few more than high-frequency band signals, for strengthening low frequency signal, the frame sign of carrying out above-mentioned reduction sampling rate needs very big.For example, frequency is the signal of 20Hz, and above-mentioned frame sign approximately is greater than 1000 samples.When above-mentioned frame sign is excessive, preserve above-mentioned sample for handling, need more internal memories, but also can take place because the echo effect (echo effect) of time delay.
Fig. 5 is the block scheme of sound intensifier according to an embodiment of the invention.
As shown in Figure 5, the sound intensifier 300 of one embodiment of the invention comprises phasing portion 310, FFT portion 320, low-frequency band enhancing portion 330, high frequency band enhancing portion 340, HRTF portion 350 and worthwhile 360.
Above-mentioned phasing portion 310 receives multichannel voice signal, for example stereophonic signal by left and right sides channel.Be the correlativity between the voice signal that reduces each channel, the above-mentioned stereophonic signal of input is carried out phasing, output to above-mentioned FFT portion 320 and above-mentioned low-frequency band enhancing portion 330 by above-mentioned left and right sides channel.Therefore above-mentioned phasing portion's 310 execution and Fig. 1, below omit its detailed description to phasing portion 110 identical functions and action shown in Figure 4.
Above-mentioned FFT portion 320 receives the voice signal of above-mentioned phasing from above-mentioned phasing portion 310 by above-mentioned left and right sides channel, voice signal to each channel carries out FFT, by each independently channel the fourier coefficient value of the low-frequency band composition of tut signal is outputed to sinusoidal wave generating unit 130, by each independently channel will output to HRTF portion 350 by all fourier coefficient values that above-mentioned FFT calculates.
Above-mentioned FFT portion 320 except above-mentioned all fourier coefficient values of being calculated by FFT being outputed to above-mentioned HRTF portion 350, carry out with Fig. 1 to FFT portion 120 identical functions and action shown in Figure 4, therefore followingly omit its detailed description.
Above-mentioned low-frequency band enhancing portion 330 receives the fourier coefficient value of the low band frequencies composition of tut signals from above-mentioned FFT portion 320 by above-mentioned each channel, generate low frequency enhancing signal and output, it comprises sinusoidal wave generating unit (not shown), overtone generating unit (not shown) and bandpass filtering portion (not shown).
Describe in detail, the sinusoidal wave generating unit of above-mentioned low-frequency band enhancing portion 330 generates the sine wave signal that has by corresponding size of the fourier coefficient value of above-mentioned each channel input and frequency, by each independently channel output to above-mentioned overtone generating unit.
Above-mentioned overtone generating unit generates the overtone signal of the sine wave signal of above-mentioned input by above-mentioned each channel classification, outputs to above-mentioned bandpass filtering portion.
Above-mentioned bandpass filtering portion receives the overtone signal by above-mentioned each channel from above-mentioned overtone generating unit, overtone signal to each channel is carried out bandpass filtering, only export the multiple that sets in advance in the overtone signal of above-mentioned each channel, the overtone signal of first and second multiple for example, the channel by each separation output to the low-frequency band portion 351 of above-mentioned HRTF portion 350.
The sinusoidal wave generating unit of above-mentioned low-frequency band enhancing portion 330, overtone generating unit and bandpass filtering portion carry out with Fig. 1 to sinusoidal wave generating unit 130, overtone generating unit 140 and bandpass filtering portion 150 identical functions and action shown in Figure 3, below omit its detailed description.
Above-mentioned high frequency band enhancing portion 340 receives the voice signals of above-mentioned execution phasing from above-mentioned phasing portion 310, generate and output high frequency enhancing signal, it comprises: first bandpass filter (not shown), reduce command troops (not shown) of sampling, frame repetition portion (not shown), partes glabra (not shown) and second bandpass filter (not shown).
The words that describe in detail, first bandpass filter of above-mentioned high frequency band enhancing portion 340 receives the voice signal of above-mentioned phasing from above-mentioned phasing portion 310, the tut signal is carried out bandpass filtering, the passband that value output sets in advance, for example the frequency content of 6KHz to 11KHz wave band outputs to above-mentioned reduction sampling afterwards and commands troops.
Above-mentioned reduction sampling commands troops the voice signal of being imported is carried out the reduction sampling rate with the frame sign that sets in advance, the multiple that sets in advance, and generates the overtone signal of tut signal.
Above-mentioned frame repetition portion carries out the frame re-treatment to the overtone signal of above-mentioned generation, and the above-mentioned overtone signal of interpolation outputs to above-mentioned second bandpass filter with the overtone signal of above-mentioned interpolation.
Above-mentioned partes glabra is carried out the smoothing processing and the output of the frame boundaries in the level and smooth above-mentioned overtone signal from the voice signal that above-mentioned frame repetition portion receives interpolation to the overtone signal of above-mentioned interpolation.
Above-mentioned second bandpass filter receives the overtone signal of above-mentioned interpolation from above-mentioned partes glabra, signal to above-mentioned interpolation is carried out bandpass filtering, only export the overtone signal of the multiple that sets in advance in the overtone signal of above-mentioned interpolation, the first overtone signal for example generates the high frequency band portion 353 that outputs to above-mentioned HRTF portion 350 after the above-mentioned high frequency enhancing signal.
First bandpass filter of above-mentioned high frequency band enhancing portion 340, reduce that sampling is commanded troops, frame repetition portion, partes glabra and second bandpass filter and Fig. 1 to first bandpass filter 220 shown in Figure 4, reduce that sampling commands troops 230, frame repetition portion 240, partes glabra 250 and second bandpass filter, 260 identical functions and action, therefore following its detailed description of omission.
350 pairs of above-mentioned low frequency enhancing signal of above-mentioned HRTF portion, the fourier transform coefficient of above-mentioned high frequency enhancing signal and tut signal are carried out the HRTF processing and are outputed to above-mentioned worthwhile 360, and it comprises, low-frequency band portion 351, main signal portion 352 and high frequency band portion 353.
The words that describe in detail, the low-frequency band portion 351 of above-mentioned HRTF portion 350 receives the overtone signal of above-mentioned bandpass filtering treatment from above-mentioned low-frequency band enhancing portion 330 by each channel, carrying out HRTF handles, it is by time lag portion and subtract the portion of declining and constitute, above-mentioned time lag portion and the subtracting portion's reflection that declines, by above-mentioned each channel, the time lag of the signal that the frequency content of above-mentioned overtone signal takes place during by spatial transmission and subtracting declines.Above-mentioned high frequency band portion 353 by a channel from killing the high frequency band enhancing portion 330 of studying and receive the overtone signal of above-mentioned bandpass filtering and carrying out HRTF and handle, by time lag portion and subtract the portion of declining and constitute, the time lag of the signal that above-mentioned time lag portion and the subtracting portion's reflection that declines, the frequency content of above-mentioned overtone signal take place during by spatial transmission and subtracting declines.Above-mentioned low-frequency band portion 351 and above-mentioned high frequency band portion's 353 each execution and Fig. 1 below omit its detailed description to HRTF portion 160 shown in Figure 3 and HRTF portion 270 identical functions and action shown in Figure 4.
Above-mentioned main signal portion 352 by each independently channel receive all fourier coefficient values of calculating by above-mentioned FFT, carrying out HRTF handles, output to above-mentioned worthwhile 360, it comprises, (the inverse fast fourier transform part: inverse fast fourier transformed portion) (not shown) of HRTF handling part (not shown) and IFFT portion.
Above-mentioned HRTF handling part receives above-mentioned fourier coefficient value, and the HRTF that carries out on the frequency field handles, and the fourier coefficient value that above-mentioned HRTF is handled outputs to above-mentioned IFFT portion.
Above-mentioned IFFT portion receives the fourier coefficient value that HRTF handles from above-mentioned HRTF, and the fourier coefficient value of above-mentioned input is carried out IFFT, will output to above-mentioned worthwhile 360 with the corresponding time zone data-voice signal of above-mentioned fourier coefficient value.
Above-mentioned worthwhile 360 from above-mentioned low-frequency band portion 351, main signal portion 352 and high frequency band portion 353 receive the voice signal that HRTF handle and carry out worthwhile after, the tut signal is outputed to loudspeaker 400 sound play devices such as grade.
Tut intensifier 300 need not to use in addition the comb filter with meticulous passband for low-frequency band strengthens, and the FFT portion 320 that is used for HRTF processing main sound signal carries out low-frequency bands and strengthens the advantage that have low complex degree, realizes easily.
Fig. 9 is that bass strengthens precedence diagram with overtone generation method according to an embodiment of the invention.
As shown in Figure 9, the phasing portion 110 that bass strengthens with overtone generating apparatus 100 receives the multichannel voice signal by left and right sides channel, stereophonic signal for example, for reducing each channel sound correlation between signals, above-mentioned stereophonic signal is carried out phasing handle, output to above-mentioned FFT portion 120 (seeing step S100) by above-mentioned left and right sides channel.
Then, the FFT portion 120 that bass strengthens with overtone generating apparatus 100 passes through the voice signal of above-mentioned left and right sides channel from the above-mentioned phasing of above-mentioned phasing portion's 110 receptions, voice signal to above-mentioned each channel is carried out Fourier transform, for example as the high speed Fourier transform of discrete Fourier transform (DFT), by each independently channel with the optional frequency composition of tut channel, the fourier coefficient value of low band frequencies composition and output to above-mentioned sinusoidal wave generating unit 130 (seeing step S110) in the time of preferably.
Afterwards, bass strengthen sinusoidal wave generating unit 130 with overtone generating apparatus 100 by above-mentioned each channel from above-mentioned each the fourier coefficient value of above-mentioned FFT portion's 120 receptions, generation has the sine wave signal with corresponding size of above-mentioned each fourier coefficient value and frequency, and the channel by each separation outputs to above-mentioned overtone generating unit 140 (seeing step S120).
Afterwards, bass strengthens the sine wave signal that receives above-mentioned generation with the set portion 135 of overtone generating apparatus 100 by above-mentioned each channel from above-mentioned sinusoidal wave generating unit 130, the sine wave signal of a part of channel in each sine wave signal is synthesized a signal, output to above-mentioned overtone generating unit 140 by a channel, the signal of all the other channels still outputs to above-mentioned overtone generating unit 140 (seeing step S130) by each channel as it is.
Afterwards, the overtone generating unit 140 that bass strengthens with overtone generating apparatus 100 receives above-mentioned each sine wave signal by above-mentioned each channel from above-mentioned sinusoidal wave generating unit 130, generation is to the overtone signal of above-mentioned each sine wave signal, by each independently channel output to above-mentioned bandpass filtering portion 150 (seeing step S140).
Afterwards, the bandpass filtering portion 150 that bass strengthens with overtone generating apparatus 100 receives the overtone signal by above-mentioned each channel from above-mentioned overtone generating unit 140, overtone signal to above-mentioned each channel is carried out bandpass filtering, only export the multiple that sets in advance in the overtone signal of above-mentioned each channel, the overtone channel of first and second multiple for example, the channel by each separation output to above-mentioned HRTF portion 160 (seeing step S150).
Afterwards, bass strengthens with the HRTF portion 160 of overtone generating apparatus 100 and receives the overtone signal of above-mentioned bandpass filtering by above-mentioned each channel, to the overtone signal of above-mentioned each channel, carries out the HRTF processing, outputs to above-mentioned worthwhile 170 (seeing step S160).
Afterwards, bass strengthen with worthwhile 170 of overtone generating apparatus 100 will be worthwhile by the overtone signal that above-mentioned each channel receives from above-mentioned HRTF portion 160 be a channel, export (seeing step S170) by a channel.
Figure 10 is that bass strengthens precedence diagram with overtone generation method according to another embodiment of the present invention.
As shown in figure 10, the phasing portion 110 that bass strengthens with overtone generating apparatus 100 receives the multichannel voice signal by left and right sides channel, stereophonic signal for example, for reducing each channel sound correlation between signals, above-mentioned stereophonic signal is carried out phasing handle, output to above-mentioned FFT portion 120 (seeing step S200) by above-mentioned left and right sides channel.
Afterwards, the low-pass filter 125 that bass strengthens with overtone generating apparatus 100 passes through the signal of above-mentioned left and right sides channel from the above-mentioned phasing of above-mentioned phasing portion's 110 receptions, voice signal to above-mentioned each channel carries out low-pass filtering, only export the ultra-low frequency signal (sub-bass frequency signal) that sets in advance, export (seeing step S210) to above-mentioned overtone generating unit 140 by above-mentioned each left and right sides channel.The preferably unlimited pulsion phase of above-mentioned low-pass filter 125 should (IIR:infinite impulse response) wave filter.
In addition, the FFT portion 120 that bass strengthens with overtone generating apparatus 100 passes through the voice signal of left and right sides channel from the above-mentioned phasing of above-mentioned phasing portion's 110 receptions, voice signal to above-mentioned each channel carries out FFT, calculates 7 fourier coefficient values to the low-frequency component of tut signal.Afterwards, above-mentioned FFT portion 120 by each independently the channel fourier coefficient that will not belong to the medium and low frequency composition (mid-bass frequency components) of above-mentioned ultralow frequency band output to above-mentioned sinusoidal wave generating unit 130 (seeing step S220).
Afterwards, bass strengthen with the sinusoidal wave generating unit 130 of overtone generating apparatus 100 by above-mentioned each independently channel from above-mentioned FFT portion 120 reception fourier coefficient values, generation has the sine wave signal of above-mentioned corresponding size of each fourier coefficient value and frequency, by each independently channel output to above-mentioned overtone generating unit 140 (seeing step S230).
Afterwards, bass strengthens that reception is above-mentioned carries out the voice signal of low-pass filtering by left and right sides channel from above-mentioned low-pass filter 125 by a channel with the overtone generating unit 140 of overtone generating apparatus 100, and generate above-mentioned input voice signal frequency content the overtone signal and output to above-mentioned bandpass filtering portion 150 by a channel.In addition, above-mentioned overtone generating unit 140 receives above-mentioned sine wave signal by above-mentioned each channel from above-mentioned sinusoidal wave generating unit 130, generation is to the overtone signal of the sine wave signal of above-mentioned each channel, by each independently channel output to above-mentioned bandpass filtering portion 150 (seeing step S240).
Afterwards, bass strengthens the input that receives the overtone signal with the bandpass filtering portion 150 of overtone generating apparatus 100 by above-mentioned each 7 channel from above-mentioned overtone generating unit 140, overtone signal to above-mentioned each channel carries out bandpass filtering, only export the multiple to set in advance in above-mentioned each channel overtone signal, the overtone signal of first and second multiple (first and second harmonics signals) for example, 7 each channel by each separation are to 160 outputs (seeing step S250) of above-mentioned HRTF portion.
Afterwards, bass strengthens with the HRTF portion 160 of overtone generating apparatus 100 and receives the overtone signal of above-mentioned bandpass filtering by above-mentioned each channel, to the overtone signal of above-mentioned each channel, carries out the HRTF processing, outputs to above-mentioned worthwhile 170 (seeing step S260).
Afterwards, bass strengthens and will be undertaken worthwhilely with worthwhile 170 of overtone generating apparatus 100 by the overtone signal that above-mentioned each channel receives from above-mentioned HRTF portion 160, exports (seeing step S270) by a channel.
Figure 11 is that bass strengthens precedence diagram with overtone generation method according to another embodiment of the present invention.
As shown in figure 11, the phasing portion 210 that high pitch strengthens with overtone generating apparatus 200 receives the multichannel voice signal by left and right sides channel, stereophonic signal for example, for reducing each channel sound correlation between signals, above-mentioned stereophonic signal is carried out phasing handle, output to above-mentioned first bandpass filter 220 (seeing step S300) by above-mentioned left and right sides channel.
Afterwards, high pitch strengthens first bandpass filter 220 with overtone generating apparatus 200 from the above-mentioned voice signal that carries out phasing of above-mentioned phasing portion's 210 receptions, the tut signal is carried out bandpass filtering, the frequency content of the passband that sets in advance of output only outputs to above-mentioned reduction sampling 230 (the seeing step S310) of commanding troops.
Afterwards, high pitch strengthens with the reduction sampling of overtone generating apparatus 200 commands troops 230 to carry out the frame re-treatment from 220 receptions of above-mentioned first bandpass filter and to the overtone signal of above-mentioned generation, outputs to above-mentioned partes glabra 250 (seeing step S320) behind the above-mentioned overtone signal of interpolation.
Afterwards, high pitch strengthens frame repetition portion 240 with overtone generating apparatus 200 in the overtone signal of above-mentioned input, fills up blank parts (seeing step S330) as the frame sign that takes place owing to above-mentioned reduction sampling rate with the voice signal sample of frame before of sampling.
Afterwards, high pitch strengthens with the partes glabra 250 of overtone generating apparatus 200 and receives the voice signal that strengthens from above-mentioned frame repetition portion 240, the overtone signal of above-mentioned interpolation is carried out smoothing processing and output, this smoothing processing is exactly the part that frame repeats in the level and smooth above-mentioned overtone signal, i.e. frame boundaries (frame edge boundary) (seeing step S340).
Afterwards, high pitch strengthens the overtone signal that receives above-mentioned interpolation with second bandpass filter 260 of overtone generating apparatus 200 from above-mentioned partes glabra 250, signal to above-mentioned interpolation is carried out bandpass filtering, only export the overtone signal of the multiple that sets in advance in the overtone signal of above-mentioned interpolation, the first overtone signal for example outputs to above-mentioned HRTF portion 270 (seeing step S350) then.Its purpose is the frequency content except that removing needed overtone signal.
Afterwards, the HRTF portion 270 that high pitch strengthens with overtone generating apparatus 200 receives above-mentioned overtone signal from above-mentioned second bandpass filter 260, the overtone signal of above-mentioned input carried out HRTF handles and output, this HRTF handles reflection sound and propagates into the variation (seeing step S360) of the above-mentioned overtone signal that takes place the way of listening taker from source of sound.
Figure 12 is that high pitch strengthens precedence diagram with overtone generation method according to an embodiment of the invention.
As shown in figure 12, the phasing portion 310 of sound intensifier 300 receives multichannel voice signal, for example stereophonic signal by left and right sides channel.Be the correlativity between the voice signal that reduces each channel, the above-mentioned stereophonic signal of input is carried out phasing, output to above-mentioned FFT portion 320 and above-mentioned low-frequency band enhancing portion 330 (seeing step S400) by above-mentioned left and right sides channel.
Afterwards, the FFT portion 320 of sound intensifier 300 is by the voice signal of above-mentioned left and right sides channel from the above-mentioned phasing of above-mentioned phasing portion's 310 receptions, voice signal to each channel carries out FFT, by each independently channel the fourier coefficient value of the low-frequency band composition of tut signal is outputed to sinusoidal wave generating unit 130, by each independently channel will output to HRTF portion 350 (seeing step S410) by all fourier coefficient values that above-mentioned FFT calculates.
Afterwards, the low-frequency band enhancing portion 330 of sound intensifier 300 from the fourier coefficient value that above-mentioned FFT portion 320 receives the low band frequencies composition of tut signals, generates low frequency enhancing signal and output (seeing step S420) by above-mentioned each channel.
Afterwards, the high frequency band enhancing portion 340 of sound intensifier 300 generates also output high frequency enhancing signal (seeing step S430) from the voice signal of the above-mentioned execution phasing of above-mentioned phasing portion's 310 receptions.
Afterwards, 350 pairs of above-mentioned low frequency enhancing signal of the HRTF portion of sound intensifier 300, the fourier transform coefficient of above-mentioned high frequency enhancing signal and tut signal is carried out HRTF and is handled and output to above-mentioned worthwhile 360 (seeing step S440).
Afterwards, worthwhile 360 of sound intensifier 300 from above-mentioned low-frequency band portion 351, main signal portion 352 and high frequency band portion 353 receive the voice signal that HRTF handle and carry out worthwhile after, the tut signal is outputed to loudspeaker 400 sound play devices such as grade (seeing step S450).
It is voice signal Play System of new generation after monophony and stereophonic sound system that 3D audio system (3D sound system) is known as.The voice signal that 3 dimensions strengthen can be given and listen taker to pass on the impression that has more presence.
The present invention is applicable to the MP3 player, the digital signal processor of the embedded platform of mobile communication terminal (embeddedplatform) or microprocessor etc.In addition, the present invention also can be used as sound and strengthens plug-in card program, is applicable to PC.The present invention is applicable to the relief music file that has more that generates the processing of 3 dimensions.
Those skilled in the art can learn by above-mentioned description, do not break away from the scope of the technology of the present invention thought to do various conversion and modification.
Thereby technical scope of the present invention is not to only limit to the content that embodiment puts down in writing, but is determined by the claim scope.
Bass of the present invention strengthens with overtone method of generationing and device and utilizes Fourier transform and sinusoidal wavely generate the comb filter that replacement has high-fineness, has that complexity is low, the effect of easy realization.
High pitch of the present invention strengthens with overtone method of generationing and device utilization and reduces the comb filter that sampling rate and the replacement of true weight compound method have high-fineness, has that complexity is low, the effect of easy realization.
Sound intensifier of the present invention uses bass to strengthen with overtone generating apparatus and high pitch enhancing and uses the overtone generating apparatus, and having to provide the effect that has more relief sound.
Above embodiment only is used to illustrate the present invention, but not is used to limit the present invention.