Classifications

• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L19/00—Speech 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
  • G10L19/02—Speech 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
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L19/00—Speech 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
  • G10L19/002—Dynamic bit allocation
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L19/00—Speech 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
  • G10L19/02—Speech 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
  • G10L19/022—Blocking, i.e. grouping of samples in time; Choice of analysis windows; Overlap factoring
  • G10L19/025—Detection of transients or attacks for time/frequency resolution switching
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
  • G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters

Definitions

• the present invention relates to encoders for encoding a signal comprising audio and / or video information, and more particularly to the estimation of a need for information units to encode that signal.
• an audio signal to be coded is fed. This is first supplied to a scaling stage 1002 in which a so-called AAC gain control is performed to set the level of the audio signal. Scaling page information is provided to a bitstream formatter 1004, as indicated by the arrow between block 1002 and block 1004. The scaled audio signal is then fed to an MDCT filter bank 1006.
• the filter bank implements a modified discrete cosine transform with 50% overlapping windows, the window length being determined by a block 1008.
• block 1008 is for windowing transient signals with shorter windows, and for windowing stationary signals with longer windows. This serves to achieve a higher time resolution (at the expense of frequency resolution) due to the shorter windows for transient signals, while for more stationary signals a higher frequency resolution (at the expense of time resolution) is achieved by longer windows is achieved, with longer windows tend to be preferred because they promise a larger Codier stand.
• temporally successive blocks of spectral values are present, which, depending on the embodiment of the filter bank, may be MDCT coefficients, Fourier coefficients or even subband signals, each subband signal having a certain limited bandwidth passing through the corresponding subband channel in the filter bank 1006, and wherein each subband signal has a certain number of subband samples.
• the filter bank outputs temporally successive blocks of MDCT spectral coefficients, which generally represent successive short-term spectra of the audio signal to be encoded at input 1000.
• a block of MDCT spectral values is then fed into a TNS processing block 1010 where temporal noise shaping (TNS) takes place.
• TNS temporal noise shaping
• the TNS technique is used to shape the temporal shape of the quantization noise within each window of the transform. This is achieved by applying a filtering process to parts of the spectral data of each channel.
• the coding is performed on a window basis. In particular, the following steps are performed to apply the TNS tool to a window of spectral data, that is, to a block of spectral values.
• a frequency range for the TNS tool is selected.
• a suitable choice is to cover a frequency range of 1.5 kHz to the highest possible scale factor band with a filter. It should be noted that this frequency range of the sampling rate depends as specified in the AAC standard (ISO / IEC 14496-3: 2001 (E)).
• LPC Linear Predictive Coding
• the expected prediction gain PG is obtained. Further, the reflection coefficients or Parcor coefficients are obtained.
• the TNS tool is not applied. In this case, control information is written in the bit stream for a decoder to know that no TNS processing has been performed.
• TNS processing is applied.
• the reflection coefficients are quantized.
• the order of the noise shaping filter used is determined by removing all the reflection coefficients having an absolute value less than a threshold from the "tail" of the reflection coefficient array. The number of remaining reflection coefficients is on the order of the noise shaping filter.
• a suitable threshold is 0.1.
• the remaining reflection coefficients are typically converted into linear prediction coefficients, which technique is also known as the N step-up- w- procedure.
• the calculated LPC coefficients are then used as coder noise shaping filter coefficients, ie as prediction filter coefficients.
• This FIR filter is routed over the specified target frequency range.
• an autoregressive filter is used, while the coding uses a so-called moving average filter.
• the page information for the TNS tool is also supplied to the bit stream formatter as shown by the arrow shown between the block TNS processing 1010 and the bitstream formatter 1004 in FIG.
• the center / side encoder 1012 is active when the audio signal to be encoded is a multi-channel signal, that is, a stereo signal having a left channel and a right channel. So far, that is, in the processing direction before the block 1012 in Fig. 3, the left and right stereo channels have been separately processed, that is, scaled, transformed by the filter bank, subjected to TNS processing or not, etc.
• middle / side encoder In the middle / side encoder is then first checked whether a middle / side encoding makes sense, that brings a coding gain at all. A middle / side encoding will then bring a coding gain if the left and the right channel are more similar, because then the center channel, that is the sum of the left and the right channel is almost equal to the left or the right channel, apart from the scaling by the factor 1/2, while the side channel has only very small values, since it is equal to the difference between the left and the right channel.
• the difference is approximately zero, or includes only very small values that are hoped to be quantized to zero in a subsequent quantizer 1014 and thus can be transmitted very efficiently, since the quantizer 1014 is followed by an entropy coder 1016.
• the quantizer 1014 is given a allowed perturbation per scale factor band by a psycho-acoustic model 1020.
• the quantizer operates iteratively, ie it first calls an outer iteration loop, which then calls an inner iteration loop.
• a quantization of a block of values is performed at the input of the quantizer 1014.
• the inner loop quantizes the MDCT coefficients, consuming a certain number of bits.
• the outer loop calculates the distortion and modified energy of the coefficients using the scale factor to again invoke an inner loop. This process is iterated until a certain conditional set is satisfied.
• the signal is reconstructed to compute the perturbation introduced by the quantization and to compare it with the allowable perturbation provided by the psycho-acoustic model 1020. Furthermore, the scale factors are increased from iteration to iteration by one step, for each iteration of the outer iteration loop.
• the analysis-through synthesis process terminates and the resulting scale factors are encoded as set forth in block 1014 and supplied in encoded form to the bitstream formatter 1004 as indicated by the arrow between block 1014 and block Block 1004 is drawn.
• the quantized values are then fed to entropy coder 1016, which typically performs entropy coding using several Huffman code tables for different scale factor bands to transfer the quantized values to a binary format.
• entropy coding in the form of Huffman coding relies on code tables that are created on the basis of expected signal statistics and in which frequently occurring values get shorter code words than more rarely occurring values.
• the entropy-coded values are then also supplied as actual main information to the bitstream formatter 1004, which then outputs the coded audio signal on the output side according to a specific bit stream syntax.
• the data reduction of audio signals is now a known technique that is the subject of a number of international standards (e.g., ISO / MPEG-1, MPEG-2 AAC, MPEG-4).
• the input signal is brought into a compact, data-reduced representation by means of a so-called encoder using perception-related effects (psychoacoustics, psychooptics).
• a spectral analysis of the signal is usually carried out and the corresponding signal components are quantized taking into account a perceptual model and subsequently coded in a compact manner as so-called bitstream.
• PE perceptual entropy
• the perceptual entropy or demand estimate of information units for encoding a signal can be used to estimate whether the signal is transient or stationary, since transient signals also require more bits to encode than more stationary signals.
• the estimation of a transient property For example, a signal is used to make a window length decision, as indicated at block 1008 in FIG. 3.
• FIG. 6 shows the perceptual entropy calculated in accordance with ISO / IEC IS 13818-7 (MPEG-2 advanced audio coding (AAC)).
• AAC MPEG-2 advanced audio coding
• the bands may originate from the band division of the psychoacoustic model (block 1020 in Fig. 3), or are the so-called scale factor bands (scfb) used in quantization.
• the psychoacoustic masking threshold is the energy value that the quantization error should not exceed.
• FIG. 6 thus shows how well such a Perceptual Entropy works as an estimate of the number of bits needed for coding.
• the respective perceptual entropy was plotted as a function of the bits consumed for each individual block using the example of an AAC coder at different bit rates.
• the test piece used contains a typical mix of music, language and individual instruments.
• the points would gather along a straight line through the zero point.
• the extension of the point sequence with the deviations from the ideal line illustrates the inaccurate estimate.
• the value for the Perceptual Entropy is determined to be too small, then the quantizer is signaled that fewer bits than actually required are needed to encode the signal. This, in turn, causes the quantizer to be coarsely quantized, which would immediately result in an audible disturbance in the signal unless countermeasures are taken.
• the countermeasures may be that the quantizer still requires one or more further iteration loops, which increases the computation time of the encoder.
• FIG. 8 shows the case in which the perceptual entropy is calculated line by line.
• the disadvantage lies in the higher computational complexity of the line-by-line calculation.
• spectral coefficients X (k) are used, where kOffset (b) designates the first index of band b.
• the object of the present invention is to provide an efficient yet accurate concept for determining an estimate of a need for information units to encode a signal.
• the present invention is based on the finding that it must be noted in a frequency band-wise calculation of the estimate for a need for information units for computing time reasons, however, that in order to obtain an accurate determination of the estimated value, the distribution the energy in the frequency band, which has to be calculated band by band.
• the entropy coder following the quantizer is implicitly "involved" in determining the estimate of the demand for information units, because entropy coding allows a smaller number of bits to be used to transmit smaller spectral values than to transmit
• the entropy coder is particularly efficient when it is possible to transmit to-zero quantized spectral values, since these will typically occur most frequently, and the codeword for transmitting a zero-quantized spectral line is the shortest codeword
• even run-length coding can be resorted to, which in the event of a run of zero On average, not even a single bit is needed per per-zero quantized spectral value.
• the measure of the distribution of energy in the frequency band can be determined based on the actual amplitudes, or by estimating the frequency lines that are not quantized to zero by the quantizer.
• This measure which is also referred to as "nl", where nl stands for “number of active lines", ie for the number of active lines, is preferred for reasons of computing efficiency.
• the number of spectral lines quantized to zero or a finer subdivision can also be taken into account, and this estimate becomes more and more accurate as more information from the downstream entropy coder is taken into account.
• the entropy coder is constructed on the basis of Huffman code tables, properties of these codetables can be integrated particularly well, since the codetables are not calculated on-line on the basis of the signal statistics, but because the codetables are fixed independently of the actual signal anyway.
• the measure of the distribution of the energy in the frequency band is determined by determining the lines still surviving after the quantization, ie the number of active lines.
• the present invention is advantageous in that an estimate of a need for information content is determined which is more accurate and more efficient than the prior art.
• the present invention is scalable for various applications since, depending on the desired accuracy of the estimate, more and more characteristics of the entropy coder, but at the cost of increased computation time, can be included in the estimation of the bit demand.
• FIG. 1 shows a block diagram of the device according to the invention for determining an estimated value
• Fig. 2a shows a preferred embodiment of the means for calculating a measure of the distribution of energy in the frequency band
• Fig. 2b shows a preferred embodiment of the means for calculating the demand for bits
• Fig. 3 is a block diagram of a known audio encoder
• FIG. 4 is a schematic diagram for explaining the influence of the energy distribution within a band on the determination of the estimated value
• 5 is a diagram for estimation calculation according to the present invention
• 6 shows a diagram for estimation calculation according to I-SO / IEC IS 13818-7 (AAC);
• the device according to the invention for determining an estimate for a requirement of information units for coding a signal is illustrated below with reference to FIG.
• the signal which may be an audio and / or a video signal, is input via an input 100.
• the signal is already present as a spectral representation with spectral values. However, this is not absolutely necessary as it can be achieved by appropriate e.g. Bandpass filtering also some calculations can be done with a time signal.
• the signal is provided to a device 102 for providing a measure of allowable interference to a frequency band of the signal.
• the allowed disturbance can be determined, for example, by means of a psycho-acoustic model, as has been explained with reference to FIG. 3 (block 1020).
• the device 102 is also operative to also provide a measure of the energy of the signal in the frequency band.
• the prerequisite for a band-wise calculation is that a frequency band for which a permitted interference or a signal energy is specified contains at least two or more spectral lines of the spectral representation of the signal.
• the frequency band will preferably be a scale factor band, since the bit demand estimate is needed directly by the quantizer to determine if a done quantization satisfies a bit criterion or not.
• the device 102 is designed to supply both the allowed disturbance nb (b) and the signal energy e (b) of the signal in the band to a device 104 for calculating the demand for bits.
• the means 104 for calculating the demand for bits is designed to take into account, in addition to the allowed disturbance and the signal energy, a measure nl (b) for a distribution of the energy in the frequency band, the distribution of the energy in the frequency band of deviates from a completely uniform distribution.
• the measure of the energy distribution is computed in a device 106, wherein the device 106 requires at least one band, namely the considered frequency band of the audio or video signal, either as a bandpass signal or directly as a series of spectral lines, e.g. to perform a spectral analysis of the band to get the measure of the distribution of energies in the frequency band.
• the audio or video signal may be supplied to the device 106 as a time signal, the device 106 then performing band filtering as well as analysis in the band.
• the audio or video signal supplied to the device 106 may already be in the frequency domain, such as MDCT coefficients, or as a bandpass signal in the filter bank with a smaller bandpass compared to an MDCT filterbank -Filter.
• the means 106 for calculating is designed to take into account current amounts of spectral values in the frequency band for calculating the estimated value.
• the means for calculating the measure of the distribution of the energy can be designed to determine as a measure of the distribution of energy a number of spectral values whose magnitude is greater than or equal to a predetermined magnitude threshold, or whose magnitude is less than or equal to the magnitude threshold wherein the magnitude threshold is preferably an estimated quantizer level that causes a quantizer to quantize values less than or equal to the quantizer level to zero.
• the measure of the energy is the number of active lines, that is, the number of lines that survive after quantization or not equal to zero.
• Fig. 2a shows a preferred embodiment of means 106 for calculating the measure of the distribution of energy in the frequency band.
• the measure of the distribution of the energy in the frequency band is designated nl (b) in FIG. 2a.
• the form factor ffac (b) is already a measure of the distribution of the energy in the frequency band.
• the measure of the spectral distribution nl from the form factor ffac (b) is weighted by the 4th root of the signal energy e (b) divided by the bandwidth width (b) and number of lines, respectively determined in the scale factor band b.
• the form factor is also an example of a quantity which gives a measure of the distribution of the energies
• nl (b) is an example of is a quantity representing an estimate of the number of lines relevant to quantization.
• the form factor ffac (b) is calculated by absolute value formation of a spectral line and subsequent rooting of this spectral line and subsequent summation of the "rooted" amounts of the spectral lines in the band.
• FIG. 2b shows a preferred embodiment of the device 104 for calculating the estimated value pe, wherein a case distinction is introduced in FIG. 2b, namely when the base 2 logarithm of the ratio of the energy to the permitted interference is greater than a constant one Factor cl or equal to the constant factor.
• the alternative above in block 104 is taken, ie the measure of the spectral distribution n1 is multiplied by the logarithm expression.
• Fig. 4a shows a band in which four spectral lines are present, all of equal size. The energy in this band is thus distributed evenly across the band.
• Fig. 4b shows a situation in which the energy in the band resides in one spectral line while the other three spectral lines are equal are zero.
• the band shown in Figure 4b could be before quantization, or could be obtained after quantization, if the spectral lines zeroed in Figure 4b are smaller than the first quantizer before quantization and thus set to zero by the quantizer So do not "survive".
• nl in Fig. 4b is calculated to the square root of 2.
• n 1 that is to say the measure for the spectral distribution of the energy in FIGS. 4 a to 4 is calculated. This means that the spectral distribution of the energy is more uniform when the measure of the distribution of the spectral energy is greater.
• the invention thus takes into account how the energy is distributed within the band. This is done as it is by replacing the number of lines per band in the known equation ( Figure 6) by an estimate of the number of lines which are non-zero after quantization. This estimate is shown in FIG. 2a.
• the form factor shown in Fig. 2a is also needed elsewhere in the encoder, for example, within the quantization block 1014 to determine the quantization step size. Then, if the form factor is already computed elsewhere, it need not be recalculated for bit estimation, so that the inventive concept of improved estimation of the measure of the required bits requires a minimum of additional computational overhead.
• X (k) is the spectral coefficient to be quantized later, while the variable k ⁇ ffset (b) designates the first index in band b.
• a measure is thus available for the characterization of the spectral field structure within the band.
• the new formula for calculating an improved band-wise perceptual entropy is thus based on the multiplication of the measure of the spectral distribution of energy and of the logarithmic expression by the signal energy e (b) in the numerator and the allowed error in the denominator, each If required, enter a term within the logarithm. can be set, as it is already shown in Fig. 7. This term may for example also be 1.5, but may also be zero, as in the case shown in Fig. 2b, this z. B. can be determined empirically.
• the method according to the invention can be implemented in hardware or in software.
• the implementation may be on a digital storage medium, in particular a floppy disk or CD with electronically readable control signals, which may interact with a programmable computer system such that the method is performed.
• the invention thus also consists in a computer program product with a program code stored on a machine-readable carrier for carrying out the method according to the invention, when the computer program product runs on a computer.
• the invention can thus be realized as a computer program with a program code for carrying out the method when the computer program runs on a computer.

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)
• Compression, Expansion, Code Conversion, And Decoders (AREA)
• Radar Systems Or Details Thereof (AREA)
• Control Of Ac Motors In General (AREA)
• Electrical Discharge Machining, Electrochemical Machining, And Combined Machining (AREA)
• Apparatus For Radiation Diagnosis (AREA)
• Measurement Of Current Or Voltage (AREA)
• Measurement Of Resistance Or Impedance (AREA)
• Branch Pipes, Bends, And The Like (AREA)
• Manufacture Or Reproduction Of Printing Formes (AREA)
• Diaphragms For Electromechanical Transducers (AREA)
• Analysing Materials By The Use Of Radiation (AREA)
• Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
• Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)

Priority Applications (4)

Applications Claiming Priority (2)

Related Child Applications (3)

Publications (2)

Family

ID=34894902

Family Applications (3)

Family Applications Before (2)

Country Status (19)

Families Citing this family (14)

Family Cites Families (22)

• 2004
  • 2004-03-01 DE DE102004009949A patent/DE102004009949B4/de not_active Expired - Fee Related
• 2005
  • 2005-02-17 PL PL08021083T patent/PL2034473T3/pl unknown
  • 2005-02-17 EP EP08021083.4A patent/EP2034473B1/fr active Active
  • 2005-02-17 KR KR1020067016835A patent/KR100852482B1/ko active IP Right Grant
  • 2005-02-17 WO PCT/EP2005/001651 patent/WO2005083680A1/fr active Application Filing
  • 2005-02-17 EP EP19167397.9A patent/EP3544003B1/fr active Active
  • 2005-02-17 CN CN2005800067994A patent/CN1938758B/zh active Active
  • 2005-02-17 ES ES05707481T patent/ES2376887T3/es active Active
  • 2005-02-17 PT PT191673979T patent/PT3544003T/pt unknown
  • 2005-02-17 CA CA2559354A patent/CA2559354C/fr active Active
  • 2005-02-17 AU AU2005217507A patent/AU2005217507B2/en active Active
  • 2005-02-17 PL PL19167397T patent/PL3544003T3/pl unknown
  • 2005-02-17 AT AT05707481T patent/ATE532173T1/de active
  • 2005-02-17 JP JP2007501149A patent/JP4673882B2/ja active Active
  • 2005-02-17 EP EP05707481A patent/EP1697931B1/fr active Active
  • 2005-02-17 PT PT08021083T patent/PT2034473T/pt unknown
  • 2005-02-17 ES ES19167397T patent/ES2847237T3/es active Active
  • 2005-02-17 BR BRPI0507815A patent/BRPI0507815B1/pt active IP Right Grant
  • 2005-02-17 DK DK05707481.7T patent/DK1697931T3/da active
  • 2005-02-17 ES ES08021083T patent/ES2739544T3/es active Active
  • 2005-02-17 RU RU2006134638/09A patent/RU2337414C2/ru active
• 2006
  • 2006-07-20 IL IL176978A patent/IL176978A/en active IP Right Grant
  • 2006-08-31 US US11/469,418 patent/US7318028B2/en active Active
  • 2006-09-29 NO NO20064432A patent/NO338917B1/no unknown
• 2007
  • 2007-01-25 HK HK07100908.4A patent/HK1093813A1/xx unknown

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events