CN101568959B - Method, medium, and apparatus with bandwidth extension encoding and/or decoding - Google Patents

Method, medium, and apparatus with bandwidth extension encoding and/or decoding Download PDF

Info

Publication number
CN101568959B
CN101568959B CN200780048069XA CN200780048069A CN101568959B CN 101568959 B CN101568959 B CN 101568959B CN 200780048069X A CN200780048069X A CN 200780048069XA CN 200780048069 A CN200780048069 A CN 200780048069A CN 101568959 B CN101568959 B CN 101568959B
Authority
CN
China
Prior art keywords
frequency
signal
tone
frequency spectrum
spectrum
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN200780048069XA
Other languages
Chinese (zh)
Other versions
CN101568959A (en
Inventor
朱基岘
吴殷美
苗磊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Samsung Electronics Co Ltd
Original Assignee
Samsung Electronics Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Samsung Electronics Co Ltd filed Critical Samsung Electronics Co Ltd
Priority claimed from PCT/KR2007/005626 external-priority patent/WO2008060068A1/en
Publication of CN101568959A publication Critical patent/CN101568959A/en
Application granted granted Critical
Publication of CN101568959B publication Critical patent/CN101568959B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
    • G10L19/02Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using spectral analysis, e.g. transform vocoders or subband vocoders
    • G10L19/0204Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using spectral analysis, e.g. transform vocoders or subband vocoders using subband decomposition
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
    • G10L19/04Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using predictive techniques
    • G10L19/06Determination or coding of the spectral characteristics, e.g. of the short-term prediction coefficients
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
    • G10L19/04Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using predictive techniques
    • G10L19/08Determination or coding of the excitation function; Determination or coding of the long-term prediction parameters
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L21/00Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
    • G10L21/02Speech enhancement, e.g. noise reduction or echo cancellation
    • G10L21/038Speech enhancement, e.g. noise reduction or echo cancellation using band spreading techniques
    • G10L21/0388Details of processing therefor

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Signal Processing (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Human Computer Interaction (AREA)
  • Acoustics & Sound (AREA)
  • Multimedia (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Quality & Reliability (AREA)
  • Compression, Expansion, Code Conversion, And Decoders (AREA)

Abstract

A method, medium, and apparatus encoding and/or decoding audio signals. By encoding and/or decoding a high-frequency signal using an excitation signal extracted from a low-frequency signal, coding efficiency can be maximized because sound quality of a signal corresponding to a high-frequency region does not deteriorate when audio signals are encoded or decoded using a low bit amounts or rates.

Description

The method of dedicated bandwidth extension encoding and/or decoding
Technical field
One or more embodiment of the present invention relates to a kind of to such as the coding audio signal of voice signal or music signal and/or method, medium and the equipment of decoding, more specifically, relate to a kind of to sound signal in the corresponding signal of high-frequency region method, medium and the equipment of encoding and/or decoding.
Background technology
Usually, the high-frequency region of sound signal typically has the identification importance that the mankind lower than corresponding low frequency region discover.Therefore, when for example emphasizing code efficiency owing to the significance bit that allows is limited, the coding of high and low frequency may on purpose cause distributing to the quantity of the position of the corresponding signal of low frequency region than distribute to the corresponding signal of high-frequency region quantity many, that is, the emphasis of coding can concentrate on low frequency region.Similarly, along with the minimizing of high-frequency region position, the transmission of the signal of the coding that obtains can have the low bit rate of signal of the coding more identical than the quantity of the position of distributing to high-frequency region and low frequency region.
Therefore, the inventor finds, when when encoding corresponding to the signal of high-frequency region, even also there is the demand of using few or the least possible figure place that method, medium and the equipment of sound quality best or that improve are provided at the discernible high-frequency region of the mankind.
Summary of the invention
Technical scheme
One or more embodiment of the present invention provides a kind of method, medium and equipment of high-frequency signal being encoded and/or decoding with the pumping signal of low frequency signal.
To be partly articulated in the following description the other aspect of the present invention and/or advantage, by describing, it can partly become clearer, perhaps can understand by implementing the present invention.
According to an aspect of the present invention, a kind of bandwidth extension encoding method is provided, described method comprises: remove envelope to extract pumping signal from low frequency signal and pumping signal is transformed to frequency domain from low frequency signal, wherein, low frequency signal belongs to the low frequency field of frequency ratio preset frequency; Produce the frequency spectrum that belongs to the high zone of frequency ratio preset frequency by the frequency spectrum of processing pumping signal; And, compare with the frequency spectrum that produces with the frequency spectrum of the high corresponding high-frequency signal in zone of frequency ratio preset frequency, and the calculated gains value.
According to a further aspect in the invention, a kind of bandwidth expansion coding/decoding method is provided, and described method comprises: remove envelope with the extraction pumping signal from low frequency signal, and pumping signal is transformed to frequency domain, wherein, low frequency signal belongs to the low frequency field of frequency ratio preset frequency; Produce the frequency spectrum that belongs to the high zone of frequency ratio preset frequency by the frequency spectrum of processing pumping signal; And, yield value is decoded, and the frequency spectrum that yield value is applied to produce.
According to a further aspect in the invention, provide a kind of bandwidth extension encoding equipment, described equipment comprises: the pumping signal extraction apparatus, remove envelope to extract pumping signal from low frequency signal, and pumping signal transformed to frequency domain, wherein, low frequency signal belongs to the low frequency field of frequency ratio preset frequency; The frequency spectrum generator produces the frequency spectrum that belongs to the high frequency field of frequency ratio preset frequency by the frequency spectrum of processing pumping signal; And the yield value counter is compared with the frequency spectrum that produces with the frequency spectrum of the high corresponding high-frequency signal in zone of frequency ratio preset frequency, and the calculated gains value.
According to a further aspect in the invention, provide a kind of bandwidth expansion decoding device, described equipment comprises: the pumping signal extraction apparatus, remove envelope to extract pumping signal from low frequency signal, and pumping signal transformed to frequency domain, wherein, low frequency signal belongs to the low frequency field of frequency ratio preset frequency; The frequency spectrum generator, the frequency spectrum of the pumping signal by processing conversion produces the frequency spectrum that belongs to the high frequency field of frequency ratio preset frequency; And the frequency spectrum applying unit is decoded to yield value, and with the yield value frequency spectrum that is applied to produce of decoding.
According to a further aspect in the invention, a kind of computer readable recording medium storing program for performing that records for the program of manner of execution is provided, described method comprises: remove envelope to extract pumping signal from low frequency signal, and pumping signal transformed to frequency domain, wherein, low frequency signal belongs to the low frequency field of frequency ratio preset frequency; Produce the frequency spectrum that belongs to the high zone of frequency ratio preset frequency by the frequency spectrum of processing pumping signal; And, compare with the frequency spectrum that produces with the frequency spectrum of the high corresponding high-frequency signal in zone of frequency ratio preset frequency, and the calculated gains value.
According to a further aspect in the invention, a kind of computer readable recording medium storing program for performing that records for the program of manner of execution is provided, described method comprises: remove envelope to extract pumping signal from low frequency signal, and pumping signal transformed to frequency domain, wherein, low frequency signal belongs to the low frequency field of frequency ratio preset frequency; Produce the frequency spectrum that belongs to the high zone of frequency ratio preset frequency by the frequency spectrum of processing pumping signal; And, yield value is decoded, and yield value should have been had the frequency spectrum that produces.
Description of drawings
By the description of embodiment being carried out below in conjunction with accompanying drawing, these and/or other aspect of the present invention and advantage will become clear and be easier to and understand, wherein:
Fig. 1 illustrates according to an embodiment of the invention bandwidth extension encoding equipment;
Fig. 2 illustrates according to an embodiment of the invention bandwidth extension encoding method;
Fig. 3 illustrates according to an embodiment of the invention bandwidth expansion decoding device;
Fig. 4 illustrates according to an embodiment of the invention bandwidth expansion coding/decoding method;
Fig. 5 illustrates according to an embodiment of the invention when the yield value that is used for four sub-frequency bands being carried out the curve that (for example, according to the decoding of the bandwidth expansion shown in Fig. 3 and Fig. 4) obtains when level and smooth;
Fig. 6 illustrates according to an embodiment of the invention (for example, according to the decoding of the bandwidth expansion shown in Fig. 3 and Fig. 4) and carries out overlapping situation.
Embodiment
Now the embodiment of the invention is described in detail, its example shown in the accompanying drawings, wherein, identical label represents same parts all the time.In this respect, embodiments of the invention can be realized by multiple different form, should not be interpreted as being limited to embodiment set forth herein.Therefore, by the reference accompanying drawing, following embodiment is only described to explain each aspect of the present invention.
Fig. 1 illustrates according to an embodiment of the invention bandwidth extension encoding equipment.At this, term " equipment " is interpreted as and term " system " synonym, be not limited to the single open or single relevant openly element of middle all descriptions of implementing in all embodiments, but, according to embodiment, can be by different elements (for example with opening, each equipment/system can be single treatment element or can be implemented by the network that distributes) can implement together or independent facilities and equipments in different open (enclosure) and/or position, additional and optional embodiment is available equally.
With reference to figure 1, for example, bandwidth extension encoding equipment can comprise: regional division unit 100, pumping signal extraction apparatus 105, the first converter unit 110, frequency spectrum generator 115, the second converter unit 120, yield value counter 125, the first tone counter 128, the second tone counter 130, tone comparer 135, yield value reduce unit 140, yield value quantizer 145, tone quantizer 150 and multiplexer 155.
For example, regional division unit 100 can (for example, by input end IN) receives signal, and is high-frequency signal and low frequency signal based on preset frequency with described division of signal.In an embodiment, low frequency signal belongs to the low frequency field of its frequency ratio the first preset frequency, and high-frequency signal belongs to the high frequency field of its frequency ratio the second preset frequency.In one embodiment, although the first preset frequency and the second preset frequency can be set to different value equally, the first preset frequency can preferably be set to identical value with the second preset frequency.
Pumping signal extraction apparatus 105 can remove envelope from (for example, obtaining from regional division unit 100) low frequency signal, therefore, extracts " pumping signal " from low frequency signal.For example, pumping signal extraction apparatus 105 can remove envelope from low frequency signal by carrying out linear predictive coding (LPC) analysis, thereby extracts pumping signal from low frequency signal.Can be approximately the hypothesis of the linear combination of the previous sample in audio samples based on audio samples, term " pumping signal " can be interpreted as the result of the forecast analysis of input signal.For example, the lpc analysis of sound signal can be attempted predicted value based on the linear combination of previous sample, and its error is value and actual currency poor of prediction.Here, then can change in lpc analysis for predicting that the linear predictor coefficient of described value is to minimize this error or optionally to produce this error.Near and final error can be outputted as " pumping signal ".By knowing linear predictor coefficient, can produce original audio signal by demoder based on the input operation inverse prediction wave filter of pumping signal.
Therefore, thus the first converter unit 110 can transform from the time domain to frequency domain with the pumping signal that obtains from low frequency signal.For example, the first converter unit 110 can be carried out Fast Fourier Transform (FFT) (FFT) to pumping signal pumping signal is transformed from the time domain to frequency domain, wherein, FFT can be for example any one among 288 FFT, 576 FFT or 1152 FFT comprise 32 288 FFT that sample is overlapping.In one embodiment, if use overlapping converter technique to be used as low frequency signal is encoded, then the first converter unit 110 can preferably use and window is set and carry out overlapping technology, thereby demoder can intactly recover low frequency signal.But the first converter unit 110 can use for the different converter technique except FFT that pumping signal is transformed from the time domain to frequency domain.For example, the first converter unit 110 can use the converter technique such as quadrature mirror filter bank (Quadrature Mirror Filterbank, QMF), in QMF in a plurality of predetermined frequency bands each is represented prearranged signals by time domain.
Frequency spectrum generator 115 can be by processing low frequency region the frequency spectrum of pumping signal of extraction in high-frequency region (for example, the high zone of frequency ratio the second preset frequency), produce frequency spectrum.For example, frequency spectrum generator 115 can produce frequency spectrum in this way in high-frequency region, namely, the frequency spectrum of the pumping signal extracted is sewed into to high-frequency region, maybe the frequency spectrum with the pumping signal extracted folds symmetrically about the example preset frequency, and this preset frequency is used for arranging separating between low frequency region and the high-frequency region.
The second converter unit 120 can transform from the time domain to frequency domain with the high-frequency signal that obtains from regional division unit 110.For example, the second converter unit 120 transforms from the time domain to frequency domain by high-frequency signal is carried out FFT with high-frequency signal, wherein, FFT can be for example any one among 288 FFT, 576 FFT or 1152 FFT comprise 32 288 FFT that sample is overlapping.In addition, in one embodiment, if use overlapping converter technique to be used as high-frequency signal is encoded, then the second converter unit 120 preferably example as window being set and carrying out overlapping technology, thereby demoder can intactly recover low frequency signal.But, it is also noted that, the second converter unit 120 can use the different converter technique except FFT that pumping signal is transformed from the time domain to frequency domain.Only as example, the second converter unit 120 can use the converter technique such as QMF, in this QMF in a plurality of predetermined frequency bands each is represented prearranged signals by time domain.
In order to obtain yield value, yield value counter 125 can also calculate by the frequency spectrum of the high-frequency signal of the second converter unit 120 conversion and the energy Ratios that is used for each predetermined frequency band in the frequency spectrum of high-frequency region that produced by frequency spectrum generator 115.
The first tone counter 128 can be the tone that unit calculates the frequency spectrum that is used for high-frequency region that is produced by frequency spectrum generator 115 by predetermined frequency band.For example, the first tone counter 128 can use spectral flatness measure (Spectral Flatness Measure, SFM) value to calculate the tone of described frequency spectrum.In an embodiment, tone becomes by deducting the value that corresponding SFM value obtains from 1.
The second tone counter 130 can be the tone that unit calculates the frequency spectrum of the high-frequency signal that is produced by the second frequency spectrum generator 120 by predetermined frequency band.
Therefore, tone comparer 135 can be compared the tone that is calculated by the first tone counter 128 with the tone that is calculated by the second tone counter.
Then, yield value reduces unit 140 can recently reduce with the tone that the second tone counter 130 calculates 125 calculating of yield value counter with respect to the energy of the tone of the first tone counter 128 calculating yield value, the large frequency band of tone of pitch ratio the first tone counter 128 calculating that the second tone counters 130 that this yield value is determined for tone comparer 135 calculate.Yield value reduces unit 140, and to reduce reason for the yield value of predetermined frequency band be to make the amount of noise of the high-frequency signal that is produced by demoder for example similar to the amount of the noise of target high-frequency signal.
Thereby, but yield value reduces unit 140 examples formula 1 described as follows and formula 2 reduces yield value.
Formula 1:
Scale = 1 - Tonality ( HB ) 1 - Tonality ( LB ) = SFM ( HB ) SFM ( LB )
At this, in this example, the tone that Tonality (HB) expression the second tone counter 130 calculates, the tone that Tonality (LB) expression the first tone counter 128 calculates, SFM (HB) expression is used for the SFM value by the frequency spectrum of the high-frequency signal of the second converter unit 120 conversion, and SFM (LB) expression is used for the SFM value by the frequency spectrum of frequency spectrum generator 115 conversion.
Formula 2:
gain′=scale×gain
At this, still in this example, gain ' expression is reduced the yield value of the predetermined frequency band that unit 140 reduces by gain, scale represents the tone that calculated by the second tone counter 130 with respect to the ratio of the tone that is calculated according to formula 1 by the first tone counter 128, and gain represents the yield value of the predetermined frequency band that calculated by yield value counter 125.
The yield value of the frequency band that yield value quantizer 145 also can reduce yield value quantizes, and described yield value reduces unit 140 by yield value and reduces.
At this, in an embodiment, the yield value that 145 pairs of yield value counters of yield value quantizer 125 calculate quantizes, this yield value is used for the little frequency band of tone that tone comparer 135 definite pitch ratio the first tone counters 128 that calculated by the second tone counter 130 calculate, that is, be used for not reducing the frequency band that unit 140 reduces by yield value.
Tone quantizer 150 can quantize the tone of each frequency band of the frequency spectrum of the high-frequency signal that is used for being calculated by the second tone counter 130.
For example, multiplexer 155 can carry out the yield value that is quantized by yield value quantizer 145 and the tone that is quantized by tone quantizer 150 multiplexing subsequently, produces bit stream, and by output terminal OUT output bit flow.
Fig. 2 illustrates according to an embodiment of the invention bandwidth extension encoding method.
At first, in operation 200, input signal can be divided into low frequency signal and high-frequency signal based on preset frequency.At this, low frequency signal can be set to belong to the frequency field that frequency is lower than the first preset frequency, and high-frequency signal can be set to belong to the frequency field that frequency is higher than the second preset frequency.According to embodiment, the first preset frequency can preferably be set to identical value with the second preset frequency, that is, and and preset frequency; But in different embodiment, first frequency and second frequency also can be set to different value.
Then, in operation 205, can from low frequency signal, remove envelope, thereby extract pumping signal from low frequency signal.Can remove envelope from low frequency signal by low frequency signal is carried out lpc analysis, thereby can extract pumping signal from low frequency signal.
Then, in operation 210, the pumping signal of low frequency signal can be able to be transformed from the time domain to frequency domain.For example, in operation 210, can use Fast Fourier Transform (FFT) (FFT), wherein, FFT can be for example any one among 288 FFT, 576 FFT or 1152 FFT comprise 32 288 FFT that sample is overlapping.In an embodiment, if use overlapping converter technique to be used as low frequency signal is encoded, then can use demoder can intactly recover window being set and carrying out overlapping technology of low frequency signal.But, in operation 210, also can use the different converter technique except FFT that time domain is transformed to frequency domain.For example, the operation 210, converter technique can be QMF, in QMF in a plurality of predetermined frequency bands each the expression time domain.
Then, in operation 215, by processing the frequency spectrum of pumping signal, can produce the frequency spectrum for the high high-frequency region of the predetermined second frequency of frequency ratio.For example, in operation 215, can produce in this way frequency spectrum, that is, the frequency spectrum of the pumping signal of the extraction that will extract from low frequency signal is sewed into to high-frequency region, and maybe the frequency spectrum with the pumping signal of extraction folds symmetrically about preset frequency.
Then, in operation 220, the high-frequency signal that obtains in the operation 200 can be transformed from the time domain to frequency domain.For example, in operation 220, be used for high-frequency signal transform to frequency domain technology can be FFT, wherein, FFT can be among 288 FFT, 576 FFT or 1152 FFT any one comprise 32 288 FFT that sample is overlapping.In an embodiment, if use overlapping converter technique to be used to high-frequency signal is encoded, then when when operation 220 execution are overlapping, using demoder can intactly recover window being set and carrying out overlapping technology of high-frequency signal.But, in operation 220, also can use the different converter technique except FFT that time domain is transformed to frequency domain.For example, in operation 220, converter technique can be the QMF technology, in QMF in a plurality of predetermined frequency bands each is represented prearranged signals by time domain.
Then, in operation 223, can be the tone that unit calculates the frequency spectrum (for example, producing in operation 220) for the high-frequency signal of conversion by predetermined frequency band.In order to calculate tone, as mentioned above, can utilize SFM.In an embodiment, in the situation that calculate tone with SFM, for example, tone can be by deducting the value that corresponding SFM value obtains from 1.
In operation 225, by each predetermined frequency band is calculated at the energy Ratios of the frequency spectrum that operates the high-frequency signal of conversion in 220 with respect to the frequency spectrum that produces, can calculate corresponding yield value in operation 215.
In addition, in operation 228, can be unit calculates the frequency spectrum that produces in operation 215 tone by predetermined frequency band.
In operation 235, also the tone that calculates in operation 228 can be compared with the tone that is used for high-frequency signal that calculates in operation 223.
Therefore, in an embodiment, in the situation of the frequency band that the tone that calculates in the pitch ratio operation 228 of the high-frequency signal that calculates in operation 223 is large, in operation 240, can be according to the tone that in operation 223, calculates with respect to reducing to operate the yield value that calculates in 225 at the ratio that operates the tone that calculates in 228.At this, similar to the amount (for example) of the noise of target high-frequency signal for the amount of the noise that makes the high-frequency signal that is produced by demoder, can operate 240 yield values that reduce for predetermined frequency band.
In operation 240, for example, can use following formula 3 and formula 4 can reduce yield value.
Formula 3:
Scale = 1 - Tonality ( HB ) 1 - Tonality ( LB ) = SFM ( HB ) SFM ( LB )
At this, Tonality (HB) is illustrated in the tone that operation 223 is calculated, Tonality (LB) is illustrated in the tone that operation 228 is calculated, and SFM (HB) expression is used for the SFM value of the frequency spectrum of high-frequency signal, and SFM (HB) expression is used for the SFM value at the frequency spectrum of operation 215.
Formula 4:
gain′=scale×gain
At this, gain ' is illustrated in the yield value of the predetermined frequency band that reduces in the operation 240, scale represents operating the ratio of 223 tones that calculate with respect to the tone that calculates by the first tone counter 128 according to formula 3 in operation 228, and gain represents by operating the yield value of 225 predetermined frequency bands that calculate.
In operation 245, can to frequency band that yield value reduce calculate at operation 240 yield values that reduce thereafter.
In the situation of the frequency band that the tone that calculates in the pitch ratio operation 228 of the high-frequency signal that calculates in operation 223 is large, can the yield value that calculate in operation 225 be quantized.
In operation 250, also can the tone at each frequency band that operates the frequency spectrum that is used for high-frequency signal that calculates in 223 be quantized.
Therefore, in operation 255, can be multiplexing by carrying out at the yield value that operates quantification in 245 and the tone that in operation 250, quantizes, also can produce synthetic bit stream.
Fig. 3 is the bandwidth extension decoder according to the embodiment of the invention.For example, with reference to figure 3, bandwidth extension decoder can comprise demodulation multiplexer 300, pumping signal extraction apparatus 305, converter 310, spectrum folding unit 315, yield value demoder 320, gain-smoothing unit 325, yield value applying unit 330, tone counter 335, tone decoder 338, tone comparer 340, noise calculation device 345, noise adder 350, inverse transformation block 355 and region synthesis device 360.
For example, demodulation multiplexer 300 can flow from the scrambler received bit by its input end, and bit stream is carried out demultiplexing.At this, demodulation multiplexer 300 can carry out demultiplexing to bit stream, with the yield value that separates each frequency band in the high zone of the frequency ratio example preset frequency comprise respectively, be used for the high zone of frequency ratio preset frequency each frequency band tone and by the low frequency signal of encoder encodes.At this, in an embodiment, low frequency signal can belong to the low zone of frequency ratio the first preset frequency, thereby corresponding high-frequency signal can be the high zone of frequency ratio the second preset frequency.In described embodiment, the first preset frequency can preferably equal the second preset frequency; But, also can the first preset frequency and the second preset frequency be set to different value.
Pumping signal extraction apparatus 305 can receive the low frequency signal of demultiplexing, and low frequency signal is decoded, and removes envelope from the low frequency signal of decoding, and extracts pumping signal from low frequency signal.At that time, pumping signal extraction apparatus 305 can extract pumping signal to remove envelope from low frequency signal by the low frequency signal of decoding is carried out lpc analysis.Therefore, pumping signal extraction apparatus 305 can extract pumping signal by the technology to extract pumping signal of using with decoded device.At this, pumping signal extraction apparatus 305 also can output to region synthesis device 360 with the low frequency signal of decoding, and the pumping signal of extracting is outputed to converter unit 310.
Converter unit 310 can transform from the time domain to frequency domain with the pumping signal of the low frequency signal that extracts.For example, converter unit 310 can transform to frequency domain with pumping signal by pumping signal is carried out FFT, and wherein, FFT can be that for example any one among 288 FFT, 576 FFT or 1152 FFT comprises 32 288 FFT that sample is overlapping.In an embodiment, if use overlapping converter technique to be used as low frequency signal is encoded, then converter unit 310 can preferably use and window is set and carry out overlapping technology, thereby demoder can intactly recover low frequency signal.But converter unit 310 can use the different converter technique except FFT that time domain is transformed to frequency domain.For example, in an embodiment, converter unit can use the converter technique such as QMF, in QMF in a plurality of predetermined frequency bands each is represented prearranged signals by time domain.
Frequency spectrum generator 315 can produce the frequency spectrum of high-frequency region, the frequency spectrum of the frequency higher than preset frequency or above-mentioned the second preset frequency by processing by the frequency spectrum of the pumping signal of converter unit 310 conversion.For example, frequency spectrum generator 315 can produce the frequency spectrum of high-frequency region in this way, that is, with extract (for example, by converter unit 310 conversion) frequency spectrum of pumping signal sews into to high-frequency region, and maybe the frequency spectrum with the pumping signal extracted folds symmetrically about preset frequency.
Yield value demoder 320 can be from the yield value of multiplexer 300 received codes, and the yield value of the coding that receives from multiplexer 300 is decoded.
In order to prevent that yield value changes sharp between frequency band, yield value smooth unit 325 also can be carried out smoothly yield value.At this, yield value smooth unit 325 can be adjusted yield value by carrying out interpolation according to the Frequency point index along the center of each frequency band between frequency band.
For example, 325 pairs of yield values that are used for four frequency bands of the smooth unit of yield value shown in Fig. 5 carry out level and smooth embodiment.Data point shown in Fig. 5 represents the yield value for four frequency bands, and the line shown in Fig. 5 represents level and smooth yield value.But in an embodiment, yield value smooth unit 325 can be not included in the bandwidth expansion decoding device.
Yield value applying unit 330 can be applied to the frequency spectrum that is produced by frequency spectrum generator 315 by (for example, level and smooth by yield value smooth unit 315) level and smooth yield value.
Tone counter 335 also can calculate the tone of having been used the frequency spectrum of yield value by yield value applying unit 330.
Tone decoder 338 can from demodulation multiplexer 300 receive high-frequency regions each frequency band tone and tone decoded, for example, this high-frequency region is corresponding to the above-mentioned high zone of the second frequency by encoder encodes of frequency ratio.
Tone comparer 340 can be compared the tone that is used for (for example, being calculated by tone counter 335) each frequency band with the tone of each frequency band that is used for being decoded by tone decoder 338.
In an embodiment, noise calculation device 345 also can be calculated to the pitch ratio that is calculated by tone counter 335 amount of the noise of tone by the large frequency band of the tone of tone decoder 338 decodings, described noise is to be used for the tone of frequency spectrum of high-frequency signal similar with the tone of being decoded by tone decoder 338.For example, the following formula 5 of noise calculation device 345 usefulness, formula 6 and formula 7 are come the amount of calculating noise.
Formula 5:
Scale LB [ i ] = Tonality ( Tag ) [ i ] Tonality ( Cur ) [ i ] = SFM ( Tag ) [ i ] SFM ( Cur ) [ i ]
Formula 6:
Scale Noise [ i ] = 1 - scale LB 2
Formula 7:
spec[j]=scale LB[i]×spec[j]+scale Loise[i]×noise[j]
At this, i represents band index, and j represents the spectrum line index.
Then, inverse transformation block 353 pitch ratio that can calculate tone counter 335 will be have for example increased the frequency spectrum of amount of noise from the frequency domain inverse transformation to time domain by noise adder 250 by the large frequency band of the tone of tone decoder decoding.For example, inverse transformation block 353 can be inverse fast Fourier transform (IFFT), wherein, for example, IFFT can be for example any one among 288 IFFT, 576 IFFT or 1152 IFFT comprise 32 288 IFFT that sample is overlapping.In an embodiment, if use overlapping converter technique to be used to low frequency signal is encoded, then inverse transformation block 353 can preferably be used and window is set and carry out overlapping technology, thereby demoder can intactly recover low frequency signal.But described inverse transformation block 353 can be used the different converter technique except IFFT that frequency domain is transformed to time domain.Only as example, inverse transformation block 353 can be used the converter technique such as QMF.
At this, therefore inverse transformation block 353 can be carried out as shown in Figure 6 overlapping.For example, if use overlapping converter technique to be used as low frequency signal is encoded, then inverse transformation block 353 can preferably be used and window is set and carry out overlapping technology, thereby demoder can intactly recover low frequency signal.
In addition, inverse transformation block 353 can have been used yield value applying unit 330 frequency spectrum of yield value from the frequency domain inverse transformation to time domain by the little frequency band of tone of tone decoder 338 decodings to the pitch ratio that is calculated by tone counter 335.
Region synthesis device 335 also can in the low zone of the above-mentioned preset frequency of frequency ratio, locate by pumping signal extraction apparatus 305 decoding low frequency signal, and in the high zone of frequency ratio example preset frequency the location by the high-frequency signal of inverse transformation block 353 inverse transformations, then low frequency signal and high-frequency signal is synthetic, and by the synthetic result of output terminal OUT output.
Fig. 4 illustrates according to an embodiment of the invention bandwidth expansion coding/decoding method.
For example, can from demoder received bit stream, then carry out demultiplexing at 400 pairs of bit streams of operation.At this, bit stream can comprise for the yield value of each frequency band in the high zone of frequency ratio preset frequency, be used for the high zone of frequency ratio preset frequency each frequency band tone and by the low frequency signal of encoder encodes.At this, in the exemplary embodiment, low frequency signal belongs to the low zone of frequency ratio the first preset frequency, thereby corresponding high-frequency signal can be the high zone of frequency ratio the second preset frequency.In described embodiment, the first preset frequency can be preferably and equal the second preset frequency; But the first preset frequency and the second preset frequency can be set to different value.
Then, in operation 405, can decode to the low frequency signal of coding, remove envelope from the low frequency signal of decoding, extract pumping signal from low frequency signal.At that time, for example, can by low frequency signal is carried out lpc analysis to remove envelope from low frequency signal, come to extract pumping signal from low frequency signal.In operation 405, can be preferably extract pumping signal by the low frequency signal of encoding with the generation that scrambler is carried out with the identical technology of technology of extracting corresponding pumping signal.
In operation 410, the pumping signal of the low frequency signal that extracts can be transformed from the time domain to frequency domain.For example, in operation 410, can use FFT, wherein, FFT can be for example any one among 288 FFT, 576 FFT or 1152 FFT comprise 32 288 FFT that sample is overlapping.In an embodiment, if use overlapping converter technique to be used to low frequency signal is encoded, then can use demoder intactly to recover window being set and carrying out overlapping technology of low frequency signal.But, in operation 410, can use the different converter technique except FFT that time domain is transformed to frequency domain.For example, in operation 410, can carry out conversion by the converter technique such as QMF, among this QMF in a plurality of predetermined frequency bands each be represented prearranged signals by time domain.
Therefore, in operation 415, can by processing the frequency spectrum of pumping signal, in the high high-frequency region of the above-mentioned preset frequency of frequency ratio (for example, the second preset frequency), produce frequency spectrum.For example, in operation 415, can produce by these modes the frequency spectrum of high-frequency region, that is, will sew into to high-frequency region at the frequency spectrum of the pumping signal of operation 410 conversion, or the frequency spectrum of pumping signal is folded into high-frequency region symmetrically about preset frequency.
Then, in operation 420, can be to being decoded by the yield value of encoder encodes.
In order to prevent that yield value from sharply changing, and in operation 425, also can carry out smoothly yield value between frequency band.At this, for example, can between frequency band, regulate yield value by carrying out interpolation according to the Frequency point index along the center of each frequency band.
For example, shown the embodiment to four level and smooth yield values of frequency band of operation 425 among Fig. 5.Data point shown in Fig. 5 represents the yield value for four frequency bands, and the line shown in Fig. 5 represents the yield value that obtains by level and smooth yield value.But, as mentioned above, in an embodiment, in the bandwidth expansion decoding technique, can not comprise described operation 425.
In operation 430, level and smooth yield value can be applied in the frequency spectrum that operation 415 produces.
In addition, in operation 435, can calculate the tone of in operation 430, having used the frequency spectrum of yield value.
In operation 438, therefore, can to as the tone for each frequency band of high-frequency region of encoder encodes decode, the frequency ratio preset frequency of this high-frequency region is high or higher than above-mentioned the second preset frequency.
Operating 440, also the tone that is used for each frequency band that calculates in operation 435 can compared with the tone that is used for each frequency band of decoding in operation 438.
In the situation of the frequency band that the tone of decoding is large in the pitch ratio operation 438 of calculating, in operation 445, calculate the amount of the tone of the frequency spectrum that the makes high-frequency signal noise similar with the tone of in operation 438, decoding in operation 435.For example, in operation 445, can be by come the amount of calculating noise with following formula 8, formula 9 and formula 10.
Formula 8:
Scale LB [ i ] = Tonality ( Tag ) [ i ] Tonality ( Cur ) [ i ] = SFM ( Tag ) [ i ] SFM ( Cur ) [ i ]
Formula 9:
Scale Noise [ i ] = 1 - scale LB 2
Formula 10:
spec[j]=scale LB[i]×spec[j]+scale Loise[i]×noise[j]
At this, i represents band index, and j represents the spectrum line index.
In operation 450, the amount of the noise that calculates in operation 445 can be applied in the frequency spectrum of having used yield value in the operation 430.
In operation 453, can large frequency band the frequency spectrum that increase the amount of noise in operation 450 be transformed to time domain from frequency domain to the tone that operates decoding in the pitch ratio operation 438 of calculating in 435.For example, in operation 453, can carry out conversion by IFFT, wherein, IFFT can be 288 the overlapping IFFT that comprise 32 samples of any one among 288 IFFT, 576 IFFT or 1152 IFFT for example.In an embodiment, if use overlapping converter technique to be used as low frequency signal is encoded, then can use demoder can intactly recover window being set and carrying out overlapping technology of low frequency signal.But, in operation 453, also can use the different converter technique except IFFT that frequency domain is transformed to time domain.For example, in operation 453, can carry out conversion by the converter technique such as QMF.
In operation 453, in an embodiment, execution that can be as shown in Figure 6 is overlapping.For example, if use overlapping converter technique to be used to low frequency signal is encoded, then can use demoder can intactly recover window being set and carrying out overlapping technology of low frequency signal.
In addition, in operation 453, to the low frequency band of tone of decoding in the pitch ratio operation 438 of in operation 435, calculating, can will in operation 430, use the frequency spectrum of yield value from the frequency domain inverse transformation to time domain.
In addition, in operation 455, by location in the low zone of the above-mentioned preset frequency of frequency ratio (for example, the operation 405 in the decoding) decoding low frequency signal and in the high zone of the described preset frequency of frequency ratio the location (for example, inverse transformation in the operation 453) high-frequency signal, low frequency signal and high-frequency signal can be carried out multiplexing, with low frequency signal and the high-frequency signal of output combination.
Except the above embodiments, also can by in recording medium (for example, computer-readable medium)/upper readable code/instruction realizes that embodiments of the invention are to realize the above-mentioned any embodiment of computing machine of at least one treatment element of control.Medium can store and/or transmit corresponding to any permission the medium/media of computer-readable code.
Record/transmission computer-readable code on medium in many ways, the example of described medium comprises: such as magnetic recording medium (for example, ROM, floppy disk, hard disk etc.) and optical recording media is (for example, CD-ROM or DVD) recording medium, such as the transmission medium that delivers or comprise the media of carrier wave, such as the Internet element.Therefore, medium can be that limit and measurable structure that comprises or deliver signal or information according to an embodiment of the invention, for example, delivers the device of bit stream.Media can also be the networks that distributes, thereby with distributed way computer-readable code are stored/transmitted and carries out.In addition, only as example, treatment element can comprise processor or computer processor, and treatment element can be distributed in and/or be included in the single element.
In the bandwidth extension encoding and/or coding/decoding method, medium and equipment according to one or more embodiment of the present invention, can be by processing from the pumping signal of low frequency signal extraction to come high-frequency signal is encoded and/or decoded.Therefore, when using a small amount of position to coding audio signal and/or decoding, the sound quality with the corresponding signal of high-frequency region is worsened, so the maximizing code efficiency.
Although specifically illustrate and described each aspect of the present invention with reference to different embodiments of the invention, should be appreciated that, these exemplary embodiments only are interpreted as descriptive rather than restrictive purpose.The ability of an aspect among any embodiment or function any dwindles or enlarges the expansion that should not be construed as similar features in different embodiment or dwindle, that is, the description of the feature among each embodiment or aspect should typically be interpreted as to other similar features among all the other embodiment or aspect can use.
Therefore, although illustrated and described some embodiment, it should be appreciated by those skilled in the art that in the situation that does not break away from the principle of the present invention that limited its scope by claim and equivalent thereof and spirit, can make amendment to these embodiment.

Claims (5)

1. bandwidth extension encoding method comprises:
Remove envelope extracting pumping signal from low frequency signal from low frequency signal, and pumping signal is transformed to frequency domain, wherein, low frequency signal belongs to the low frequency field of frequency ratio preset frequency;
Produce the frequency spectrum that belongs to the high zone of frequency ratio preset frequency by the frequency spectrum of processing pumping signal; With
Compare with the frequency spectrum that produces with the frequency spectrum of the high corresponding high-frequency signal in zone of frequency ratio preset frequency, and calculate the yield value of the frequency spectrum of high-frequency signal;
Use the spectral flatness measure value to calculate the tone of the frequency spectrum of the tone of the frequency spectrum that produces and high-frequency signal, and the tone of the frequency spectrum that produces is compared with the tone of the frequency spectrum of high-frequency signal;
Adjust yield value according to result relatively.
2. the step that the method for claim 1, wherein pumping signal is transformed to frequency domain comprises: come to extract pumping signal from low frequency signal to remove envelope from low frequency signal by low frequency signal being carried out linear forecast coding analysis.
3. the method for claim 1, wherein, the step that produces the frequency spectrum belong to the high zone of frequency ratio preset frequency comprises: the frequency spectrum by the pumping signal that will extract from low frequency signal folds into symmetrically the frequency spectrum of the pumping signal that maybe will extract from low frequency signal in the high zone of frequency ratio preset frequency and sews into producing frequency spectrum to the high zone of frequency ratio preset frequency.
4. the method for claim 1 also comprises: yield value and tone to the frequency spectrum of high-frequency signal when the tone of the frequency spectrum that produces is not more than the tone of frequency spectrum of high-frequency signal are encoded,
When the tone of the frequency spectrum that produces during greater than the tone of the frequency spectrum of high-frequency signal yield value and the tone to the adjustment of the frequency spectrum of high-frequency signal encode.
5. the method for claim 1, wherein the step of calculated gains value comprises: calculate high-frequency signal frequency spectrum energy value with respect to for generation of the ratio of energy value of frequency spectrum, thereby the calculated gains value.
CN200780048069XA 2006-11-17 2007-11-08 Method, medium, and apparatus with bandwidth extension encoding and/or decoding Expired - Fee Related CN101568959B (en)

Applications Claiming Priority (7)

Application Number Priority Date Filing Date Title
KR10-2006-0114101 2006-11-17
KR1020060114101 2006-11-17
KR20060114101 2006-11-17
KR10-2007-0046203 2007-05-11
KR1020070046203A KR101375582B1 (en) 2006-11-17 2007-05-11 Method and apparatus for bandwidth extension encoding and decoding
KR1020070046203 2007-05-11
PCT/KR2007/005626 WO2008060068A1 (en) 2006-11-17 2007-11-08 Method, medium, and apparatus with bandwidth extension encoding and/or decoding

Publications (2)

Publication Number Publication Date
CN101568959A CN101568959A (en) 2009-10-28
CN101568959B true CN101568959B (en) 2013-10-23

Family

ID=39662686

Family Applications (1)

Application Number Title Priority Date Filing Date
CN200780048069XA Expired - Fee Related CN101568959B (en) 2006-11-17 2007-11-08 Method, medium, and apparatus with bandwidth extension encoding and/or decoding

Country Status (2)

Country Link
KR (2) KR101375582B1 (en)
CN (1) CN101568959B (en)

Families Citing this family (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9294060B2 (en) * 2010-05-25 2016-03-22 Nokia Technologies Oy Bandwidth extender
US8600737B2 (en) * 2010-06-01 2013-12-03 Qualcomm Incorporated Systems, methods, apparatus, and computer program products for wideband speech coding
JP6075743B2 (en) * 2010-08-03 2017-02-08 ソニー株式会社 Signal processing apparatus and method, and program
KR101826331B1 (en) * 2010-09-15 2018-03-22 삼성전자주식회사 Apparatus and method for encoding and decoding for high frequency bandwidth extension
BR112013033900B1 (en) * 2011-06-30 2022-03-15 Samsung Electronics Co., Ltd Method to generate an extended bandwidth signal for audio decoding
WO2013141638A1 (en) * 2012-03-21 2013-09-26 삼성전자 주식회사 Method and apparatus for high-frequency encoding/decoding for bandwidth extension
CN103971694B (en) * 2013-01-29 2016-12-28 华为技术有限公司 The Forecasting Methodology of bandwidth expansion band signal, decoding device
CN104517611B (en) * 2013-09-26 2016-05-25 华为技术有限公司 A kind of high-frequency excitation signal Forecasting Methodology and device
US10163447B2 (en) * 2013-12-16 2018-12-25 Qualcomm Incorporated High-band signal modeling
JP6383000B2 (en) 2014-03-03 2018-08-29 サムスン エレクトロニクス カンパニー リミテッド High frequency decoding method and apparatus for bandwidth extension
KR102386736B1 (en) * 2014-03-03 2022-04-14 삼성전자주식회사 Method and apparatus for decoding high frequency for bandwidth extension
CN106463133B (en) 2014-03-24 2020-03-24 三星电子株式会社 High-frequency band encoding method and apparatus, and high-frequency band decoding method and apparatus
CN112885363A (en) * 2019-11-29 2021-06-01 北京三星通信技术研究有限公司 Voice sending method and device, voice receiving method and device and electronic equipment

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1397064A (en) * 2000-01-28 2003-02-12 艾利森电话股份有限公司 System and method for modifying speech signals

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE602006012637D1 (en) * 2005-04-01 2010-04-15 Qualcomm Inc Apparatus and method for subband speech coding

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1397064A (en) * 2000-01-28 2003-02-12 艾利森电话股份有限公司 System and method for modifying speech signals

Also Published As

Publication number Publication date
KR101375582B1 (en) 2014-03-20
KR20080045047A (en) 2008-05-22
CN101568959A (en) 2009-10-28
KR101376098B1 (en) 2014-03-31
KR20130112817A (en) 2013-10-14

Similar Documents

Publication Publication Date Title
CN101568959B (en) Method, medium, and apparatus with bandwidth extension encoding and/or decoding
US8639500B2 (en) Method, medium, and apparatus with bandwidth extension encoding and/or decoding
KR101747918B1 (en) Method and apparatus for decoding high frequency signal
EP2056294B1 (en) Apparatus, Medium and Method to Encode and Decode High Frequency Signal
CN102576542B (en) Method and device for determining upperband signal from narrowband signal
CN101518083B (en) Method, medium, and system encoding and/or decoding audio signals by using bandwidth extension and stereo coding
JP4918841B2 (en) Encoding system
KR100949232B1 (en) Encoding device, decoding device and methods thereof
US8861738B2 (en) Method, medium, and system encoding/decoding multi-channel signal
US9177569B2 (en) Apparatus, medium and method to encode and decode high frequency signal
JP4272897B2 (en) Encoding apparatus, decoding apparatus and method thereof
US20080077412A1 (en) Method, medium, and system encoding and/or decoding audio signals by using bandwidth extension and stereo coding
KR20110095354A (en) Audio encoder and bandwidth extension decoder
CN101542599A (en) Method, apparatus, and system for encoding and decoding broadband voice signal
US20060251178A1 (en) Encoder apparatus and decoder apparatus
EP3179476B1 (en) Coding device and method, and program
US20120123788A1 (en) Coding method, decoding method, and device and program using the methods
CN105556602A (en) Frequency band table design for high frequency reconstruction algorithms
JP3094522B2 (en) Vector quantization method and apparatus

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20131023

Termination date: 20201108

CF01 Termination of patent right due to non-payment of annual fee