WO2011000434A1 - Dispositif - Google Patents

Dispositif Download PDF

Info

Publication number
WO2011000434A1
WO2011000434A1 PCT/EP2009/058438 EP2009058438W WO2011000434A1 WO 2011000434 A1 WO2011000434 A1 WO 2011000434A1 EP 2009058438 W EP2009058438 W EP 2009058438W WO 2011000434 A1 WO2011000434 A1 WO 2011000434A1
Authority
WO
WIPO (PCT)
Prior art keywords
coefficient
interval
value
determined threshold
threshold
Prior art date
Application number
PCT/EP2009/058438
Other languages
English (en)
Inventor
Florin Ghido
Original Assignee
Nokia Corporation
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 Nokia Corporation filed Critical Nokia Corporation
Priority to PCT/EP2009/058438 priority Critical patent/WO2011000434A1/fr
Publication of WO2011000434A1 publication Critical patent/WO2011000434A1/fr

Links

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
    • G10L19/04—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 predictive techniques
    • G10L19/06—Determination or coding of the spectral characteristics, e.g. of the short-term prediction coefficients
    • 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
    • G10L19/0017—Lossless audio signal coding; Perfect reconstruction of coded audio signal by transmission of coding error
    • 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
    • G10L19/032—Quantisation or dequantisation of spectral components
    • G10L19/035—Scalar quantisation

Definitions

  • the present application relates to apparatus for coding, and in particular, but not exclusively to apparatus for the quantization of coefficients associated with speech or audio coding.
  • Audio signals like speech or music, are encoded for example for enabling an efficient transmission or storage of the audio signals.
  • Audio encoders and decoders are used to represent audio based signals, such as music and background noise. These types of coders typically do not utilise a speech model for the coding process, rather they use processes for representing all types of audio signals, including speech.
  • Speech encoders and decoders are usually optimised for speech signals, and can operate at either a fixed or variable bit rate.
  • Many speech and audio coders adopt the technique of Linear Prediction Coding (LPC) in order to efficiently represent the source signal. This technique relies on the audio signal exhibiting a significant amount of correlation between successive samples. These correlations may be known as short term correlations whereby LPC analysis attempts to model these correlations in the form of a very efficient short order linear autoregressive filter.
  • LPC Linear Prediction Coding
  • a typical LPC analysis based coder may partition the samples of the input audio signal into a number of frames. The length of each frame may be determined by the sampling rate of the input audio signal and the requirement that the signal is considered to be quasi-stationary for the duration of the frame. LPC analysis may then be performed on each frame in order to determine the LPC coefficients for the linear autoregressive filter. The linear autoregressive filter may then be used to model the short term correlations between the samples for each frame.
  • the encoding process typically involves quantizing the LPC coefficients. However, LPC coefficients are particularly susceptible to quantization errors or noise which can have the effect of causing the linear autoregressive filter to become unstable.
  • LPC coefficients are typically transformed into a different format which is more suited to the process of quantization.
  • Typical examples of formats which may be used to represent LPC coefficients include, Log Area Ratios (LAR), Reflection coefficients and Line Spectral Frequencies (LSF).
  • reflection coefficient representation of the LPC coefficient is used in coding systems where modelling in the time domain is considered more important than modelling in the spectral domain.
  • a further advantage of using reflection coefficients pertains from the fact that these coefficients may be determined directly from the intermediary steps of the Levinson-Durbin algorithm, which is used to calculate the LPC coefficients.
  • Reflection coefficients exhibit a number of properties which make them particularly suitable for quantization. For instance, as mentioned above they may be determined with minimum overhead since they are produced as a by product of the Levinson-Durbin algorithm. Additionally reflection coefficients are bounded between the range +1 to -1 which not only facilitates the process of quantization but also provides for a simple stability check once the coefficients are dequantized. Finally, since reflection coefficients may be efficiently scalar quantized they tend to be more robust to quantization errors than LPC quantization schemes which adopt vector quantization techniques. However, reflection coefficients are particularly sensitive to quantization noise for values in the region of +1 and -1 since these values typically coincide with regions of high coding gain in the decoded signal. For instance, the effect of quantization noise in this region may have a profound effect on the shape of the decoded waveform.
  • the sensitivity of reflection coefficients to quantization noise may be compensated for by employing non-uniform quantization techniques, whereby proportionally more of the quantization "effort" is directed to the more sensitive reflection coefficient values. For instance such techniques may result in proportionally more quantization levels being allocated for reflection coefficients values in the region of +1 and -1 than in other regions.
  • One technique which may be used to non uniformly quantize reflection coefficients comprises scaling the reflection coefficient by a non linear companding function.
  • the companding function has the effect of enlarging the quantization space for the more sensitive reflection coefficient values (i.e. those values around + and -1 ) and decreasing the quantization space associated with the least sensitive coefficient values.
  • the resulting companded reflection coefficient value may then be quantized with a uniform quantizer. The effect of this technique is to place a higher emphasis on the more sensitive reflection coefficient values during the quantization process, thereby resulting in a smaller quantization error for these values.
  • Embodiments of the present application aim to address the above problem.
  • a method comprising: mapping a coefficient to an interval of an uniform quantization scale; determining whether a value of the coefficient satisfies corresponds to at least one of at least one p re-determined threshold; and further mapping the coefficient to a non-uniform quantization scale dependant on whether the value of the coefficient satisfies at least one of the at least one p re-determined threshold.
  • the at least one p re-determined threshold may correspond to a threshold interval of the uniform quantization scale, wherein each of the at least one pre- determined threshold interval may be associated with at least one refinement bit.
  • the method may further comprise dividing the interval into which the coefficient is mapped into a plurality of sub-intervals, wherein an index is assigned to each sub-interval of the plurality of sub-intervals; determining a sub-interval from the plurality of sub-intervals whose numerical range spans the value of the coefficient; selecting the sub-interval whose numerical range spans the value of the coefficient; and associating the index of the selected sub-interval with the coefficient.
  • the number of plurality of sub-intervals maybe determined by an at least one refinement bit numerical value associated with the at least one pre-determined threshold value.
  • the method may further comprise dividing the interval into which the coefficient is mapped into a plurality of equally proportioned sub-intervals.
  • the method may further comprise determining whether the interval into which the coefficient is mapped is equal to or exceeds at least one of the at least one threshold interval.
  • the method may further comprise determining whether the interval into which the coefficient is mapped is equal to or less than at least one of the at least one threshold interval.
  • the at least one pre-determined threshold may be dependant on a rate distortion value associated with a mean square error signal.
  • the method may further comprise determining the mean square error by a difference between a first predicted signal and a second predicted signal.
  • the prediction order for the first predicted signal may be greater than the prediction order of the second predicted signal.
  • the rate distortion value may be associated with at least one equivalent fractional bit.
  • the at least one refinement bit associated with each of the at least one predetermined threshold interval may be determined from the at least one equivalent fractional bit from the associated rate distortion value.
  • the at least one p re-determined threshold may be determined by calculating the rate distortion value associated with the at least one equivalent fractional bit.
  • the method may further comprise determining a uniform quantization interval from the uniform quantization scale whose numerical range spans the value of the coefficient; selecting the uniform quantization interval whose numerical range spans the value of the coefficient; and associating an index of the selected uniform quantization interval with the coefficient.
  • the method may further comprise representing the coefficient at least in part by the index associated with the selected sub-interval and the index associated with the selected uniform quantization interval.
  • the at least one pre-determined threshold may comprise a first pre-determined threshold and a second pre-determined threshold
  • the first pre-determined threshold may have a value of 0.7071 for positive value coefficients and may have a value of -0.7071 for negative value coefficients.
  • the second pre-determined threshold may have a value of 0.9354 for a positive value coefficient and may have value of -0.9354 for a negative value coefficients, where one refinement bit may be associated with a threshold interval associated with the first of the at least one pre-determined threshold, and where two refinement bits may be associated with a threshold interval associated with the second of the at least one pre-determined threshold.
  • the value of the coefficient may lie within the range from -1 to +1.
  • the coefficient may be a reflection coefficient.
  • an apparatus comprising at least one processor and at least one memory including computer program code, the at least one memory and the computer program code configured to with the at least one processor cause the apparatus to perform: mapping a coefficient to an interval of an uniform quantization scale; determining whether a value of the coefficient satisfies at least one of at least one pre-determined threshold; and further mapping the coefficient to a nonuniform quantization scale dependant on whether the value of the coefficient satisfies at least one of the at least one pre-determined threshold.
  • the at least one pre-determined threshold may correspond to a threshold interval of the uniform quantization scale, wherein each of the at least one predetermined threshold interval is associated with at least one refinement bit.
  • the at least one processor and at least one memory may be further configured to perform: dividing the interval into which the coefficient is mapped into a plurality of sub-intervals, wherein an index is assigned to each sub-interval of the plurality of sub-intervals; determining a sub-interval from the plurality of sub- intervals whose numerical range spans the value of the coefficient select the sub-interval whose numerical range spans the value of the coefficient; and associating the index of the selected sub-interval with the coefficient.
  • the number of the plurality of sub-intervals may be determined by an at least one refinement bit numerical value associated with the at least one pre determined threshold interval.
  • the at least one processor and at least one memory may be further configured to perform dividing the interval into which the coefficient is mapped into a plurality of equally proportioned sub-intervals.
  • the at least one processor and at least one memory may be further configured to perform determining whether the interval into which the coefficient is mapped is equal to or exceeds at least one of the at least one threshold interval.
  • the at least one processor and at least one memory may be further configured to perform determining whether the interval into which the coefficient is mapped is equal to or is less than at least one of the at least one threshold interval.
  • the at least one pre-determined threshold is preferably dependant on a rate distortion value associated with a mean square error signal.
  • the at least one processor and at least one memory may be further configured to perform determining the mean square error by a difference between a first predicted signal and a second predicted signal.
  • the prediction order for the first predicted signal may be greater than the prediction order of the second predicted signal.
  • the rate distortion value may be associated with at least one equivalent fractional bit.
  • the at least one refinement bit associated with each of the at least one pre- determined threshold interval may be determined from the at least one equivalent fractional bit from the associated rate distortion value.
  • the at least one pre-determined threshold may be determined by calculating the rate distortion value associated with the at least one equivalent fractional bit.
  • the at least one processor and at least one memory may be further configured to perform determining a uniform quantization interval from the uniform quantization scale whose numerical range spans the value of the coefficient; select the uniform quantization interval whose numerical range spans the value of the coefficient; and associate an index of the selected uniform quantization interval with the coefficient.
  • the at least one processor and at least one memory may be further configured to perform representing the coefficient at least in part by the index associated with the selected sub-interval and the index associated with the selected uniform quantization interval.
  • the at least one pre-determined threshold may comprise a first pre-determined threshold and a second pre-determined threshold
  • the first pre-determined threshold may have a value of 0.7071 for positive value coefficients and may have a value of -0.7071 for negative value coefficients.
  • the second pre-determined threshold may have a value of 0.9354 for a positive value coefficient and may have value of -0.9354 for a negative value coefficients, where one refinement bit may be associated with a threshold interval associated with the first of the at least one pre-determined threshold, and where two refinement bits may be associated with a threshold interval associated with the second of the at least one pre-determined threshold.
  • the value of the coefficient may lie within the range from -1 to +1.
  • the coefficient may be a reflection coefficient.
  • a computer-readable medium encoded with instructions that, when executed by a computer, perform mapping a coefficient to an interval of an uniform quantization scale; determining whether a value of the coefficient satisfies at least one of at least one pre-determined threshold; and further mapping the coefficient to a non-uniform quantization scale dependant on whether the value of the coefficient satisfies at least one of the at least one pre-determined threshold.
  • an apparatus comprising means for mapping a coefficient to an interval of an uniform quantization scale; means for determining whether a value of the coefficient satisfies at least one of at least one pre-determined threshold; and means for further mapping the coefficient to a non-uniform quantization scale dependant on whether the value of the coefficient satisfies at least one of the at least one pre-determined threshold.
  • An electronic device may comprise apparatus as described above.
  • a chipset may comprise apparatus as described above.
  • an apparatus comprising: a uniform quantizer configured to map a coefficient to an interval of an uniform quantization scale; an interval threshold determiner configured to determine whether a value of the coefficient satisfies at least one of at least one pre-determined threshold; and a non-uniform quantizer configured to further map the coefficient to a non-uniform quantization scale dependant on the value of the coefficient satisfying at least one of the at least one pre-determined threshold.
  • FIG 1 shows schematically an electronic device employing some embodiments of the invention
  • FIG. 2 shows schematically an audio codec system employing some embodiments of the present invention
  • Figure 3 shows schematically an encoder part of the audio codec system shown in figure 2;
  • Figure 4 shows an example of a uniform quantization scale as employed in some embodiments of the invention; schematically a decoder part of the audio codec system shown in figure 2;
  • Figure 5 shows a flow diagram illustrating the operation of an embodiment as shown in Figure 3 according to the present application.
  • Figure 6 shows schematically a decoder part of the audio codec system shown in figure 2.
  • FIG. 1 schematic block diagram of an exemplary electronic device 10, which may incorporate a codec according to an embodiment of the invention.
  • the electronic device 10 may for example be a mobile terminal or user equipment of a wireless communication system.
  • the electronic device 10 comprises a microphone 11 , which is linked via an analogue-to-digital converter 14 to a processor 21.
  • the processor 21 is further linked via a digital-to-analogue converter 32 to loudspeakers 33.
  • the processor 21 is further linked to a transceiver (TX/RX) 13, to a user interface (Ul) 15 and to a memory 22.
  • the processor 21 may be configured to execute various program codes.
  • the implemented program codes comprise an audio encoding code for encoding a lower frequency band of an audio signal and a higher frequency band of an audio signal.
  • the implemented program codes 23 further comprise an audio decoding code.
  • the implemented program codes 23 may be stored for example in the memory 22 for retrieval by the processor 21 whenever needed.
  • the memory 22 could further provide a section 24 for storing data, for example data that has been encoded in accordance with the invention.
  • the encoding and decoding code may in embodiments of the invention be implemented in hardware or firmware.
  • the user interface 15 enables a user to input commands to the electronic device 10, for example via a keypad, and/or to obtain information from the electronic device 10, for example via a display.
  • the transceiver 13 enables a communication with other electronic devices, for example via a wireless communication network.
  • a user of the electronic device 10 may use the microphone 11 for inputting speech that is to be transmitted to some other electronic device or that is to be stored in the data section 24 of the memory 22.
  • a corresponding application has been activated to this end by the user via the user interface 15. This application, which may be run by the processor 21 , causes the processor 21 to execute the encoding code stored in the memory 22.
  • the analogue-to-digital converter 14 converts the input analogue audio signal into a digital audio signal and provides the digital audio signal to the processor 21.
  • the processor 21 may then process the digital audio signal in the same way as described with reference to Figures 2 and 3.
  • the resulting bit stream is provided to the transceiver 13 for transmission to another electronic device.
  • the coded data could be stored in the data section 24 of the memory 22, for instance for a later transmission or for a later presentation by the same electronic device 10.
  • the electronic device 10 could also receive a bit stream with correspondingly encoded data from another electronic device via its transceiver 13.
  • the processor 21 may execute the decoding program code stored in the memory 22.
  • the processor 21 decodes the received data, and provides the decoded data to the digital-to-analogue converter 32.
  • the digital-to-analogue converter 32 converts the digital decoded data into analogue audio data and outputs them via the loudspeakers 33. Execution of the decoding program code could be triggered as well by an application that has been called by the user via the user interface 15. The received encoded data could also be stored instead of an immediate presentation via the loudspeakers 33 in the data section 24 of the memory 22, for instance for enabling a later presentation or a forwarding to still another electronic device.
  • FIG. 1 The general operation of audio codecs as employed by embodiments of the invention is shown in figure 2.
  • General audio coding/decoding systems consist of an encoder and a decoder, as illustrated schematically in figure 2. Illustrated is a system 102 with an encoder 104, a storage or media channel 106 and a decoder 108.
  • the encoder 104 compresses an input audio signal 1 10 producing a bit stream 112, which is either stored or transmitted through a media channel 106.
  • the bit stream 112 can be received within the decoder 108.
  • the decoder 108 decompresses the bit stream 1 12 and produces an output audio signal 114.
  • the bit rate of the bit stream 112 and the quality of the output audio signal 1 14 in relation to the input signal 1 10 are the main features, which define the performance of the coding system 102.
  • Figure 3 shows schematically an encoder 104 according to an embodiment of the invention.
  • the encoder 104 comprises an input 203 arranged to receive an audio signal.
  • the input 203 may be arranged to receive an audio signal which is Pulse Code Modulated (PCM) encoded to the International Telecommunications Union Standard (ITU) G.711.
  • PCM Pulse Code Modulated
  • ITU International Telecommunications Union Standard
  • the input signal may be a companded PCM signal whereby the companding may be according to either A-Law or ⁇ -Law compression standard.
  • the input 203 may be connected to a decompander 210.
  • the decompander 210 may apply the inverse of the compression function used to compand the input PCM audio signal.
  • the inverse of the compression function as determined by the particular compression standard maybe performed on a sample by sample basis by the decompander 210.
  • the input 203 may be arranged to receive an audio signal which is a linear PCM signal. Therefore, it is to be understood in such embodiments that there may not be a decompander 210 at the input to the encoder 104.
  • the output from the decompander 210 may be connected to the input of the frame collector 212.
  • the frame collector 212 may collate a plurality of input PCM samples in order to form a frame of audio samples.
  • the frame collector 212 may collate the input audio samples into a frame size of one of 40, 80, 160, 240 or 320 samples.
  • the number of samples in each frame may be determined by the mode of operation of the codec. For example, if the encoder 104 is selected to operate with a frame size of 40 samples then all subsequent samples will be collated into frames of 40 samples.
  • the frame collector 212 may apply a window function to the frame of audio samples.
  • the window function applied to the frame of audio samples may comprise any one of the following types; Hamming, Hanning, Kaiser, Blackman or Bartlett.
  • the output from the frame collector 212 may be connected to the input of the optimal predictor order estimator 214.
  • the optimal predictor order estimator 214 may determine an optimal filter order for the subsequent optimal predictor calculator 216. In other words the optimal predictor order estimator determines from the input audio frame the optimal number of filter coefficients which may be used by the subsequent optimal predictor calculator 216 and predictor 220. In some embodiments the optimal predictor order estimator 214 may determine optimal filter order by the size of the input audio frame. In these embodiments this may be implemented by predetermining a filter order for each input audio frame sizes. For example, an audio frame size of 40 samples may be pre-allocated a lower predictor order, whereas a frame size of 320 samples may be pre-allocated a higher predictor order.
  • the optimal filter order estimation criteria may be further enhanced by including further information relating to the type of companding used by the audio signal at the input to the encoder 104.
  • the optimal predictor order estimation criteria may be further enhanced by incorporating information about the amplitude of the audio signal in the companded domain. For instance, if the audio signal is of low amplitude this may indicate that the signal is noise like thereby exhibiting relatively low level of correlation. This may bias the decision of the optimal predictor order estimator to select a lower predictor order.
  • the optimal predictor order estimator may be biased to select a higher predictor order.
  • the criteria used by the optimal predictor order estimator may be an amalgamation of any of the above.
  • the optimal predictor order estimator 214 may be implemented as a pre-computed look-up table stored in memory.
  • the filter order estimate output from the optimal predictor order estimator 214 may be connected to the input of the optimal predictor calculator 216. Additionally the optimal predictor calculator 216 may be arranged to receive a further input comprising the frame of audio samples from the frame collector 212.
  • the optimal predictor calculator 216 analyses the input audio frame in order to determine the prediction or filter coefficients for the subsequent predictor 220. As mentioned above the number of filter coefficients to be calculated is determined by the filter order estimate output from the optimal predictor order estimator 214.
  • the optimal predictor calculator 216 may determine the filter coefficients by analysing the short term correlations in the audio frame as provided by the frame collector 212.
  • the analysis of the short term correlations of the audio frame may be accomplished by linear predictive coding (LPC) analysis.
  • LPC linear predictive coding
  • This technique relies on either calculating the autocovariance or autocorrelation of the audio frame over a range of different sample delays, whereby the range of sample delays may be typically determined by the number filter order.
  • the LPC analysis may be performed using the autocorrelation method whereby the result of calculating the autocorrelations over the range of different delays as determined by the filter order may be formed into a symmetrical square matrix known as a Toeplitz matrix.
  • the Toeplitz matrix has the property that it is symmetrical about the leading diagonal and all the elements along any given diagonal are equal.
  • the matrix may be inverted using the Levinson-Durbin algorithm.
  • the corresponding reflection coefficients are generated as an intermediary step during the calculation process for the LPC coefficients when using the Levinson-Durbin algorithm.
  • the LPC analysis may be performed using the autocovariance method.
  • the covariance over the range of different delays of samples within the audio frame may be determined in order to form a covariance matrix.
  • the size of the matrix is determined by the range of delays over which the various values of covariance are calculated.
  • the range of delays over which the values of the covariance are calculated are determined by the number of LPC coefficients and hence the order of the subsequent LPC filter 220.
  • the covariance matrix is symmetrical about the leading diagonal. However, unlike the Toeplitz matrix the values within a given diagonal are not necessary equal. In these embodiments the matrix may be inverted using Cholesky Decomposition in order to derive the LPC coefficients.
  • the corresponding reflection coefficients are not produced as an intermediary step within the Cholesky Decomposition algorithm. In such embodiments it may therefore be required to have an additional conversion step which converts the set of calculated LPC coefficients into their corresponding set of reflection coefficients. It is to be further understood in these second groups of embodiments that unlike the autocorrelation method, the covariance method does not require that the audio frame is scaled with a suitable windowing function before LPC analysis. Consequently in such embodiments the windowing functionality within the frame collector 212 may not be performed.
  • the output from the optimal predictor designer 216 is the set of reflection coefficients corresponding to the LPC coefficients for the subsequent predictor 220.
  • the output from the optimal predictor designer 216 may be connected to the input to the reflection coefficient quantizer 218.
  • the reflection coefficient quantizer 218 may then quantize each one of the set of reflection coefficients on a scalar basis.
  • each reflection coefficient may be performed using a nonuniform scalar quantization method which may be an extension of a uniform quantization method.
  • each reflection coefficient may be uniformly scalar quantized and then depending on an estimation criterion the uniform scalar quantized reflection coefficient may be further quantized in accordance with a nonuniform quantization scale.
  • each reflection coefficient may be performed on an individual basis. It is to be appreciated that the value of a reflection coefficient is bound between the values +1 to -1. In other words a reflection coefficient may have any value within the range [-1 , +1]. It is to be further appreciated that this property or reflection coefficients makes them particularly suitable for quantization.
  • a reflection coefficient may be scalar quantized according to a particular uniform scale.
  • the number of quantization levels within the uniform scale may be determined by the number of bits available. For example, if three bits are available for uniform quantization of the reflection coefficient then there may be up to 2 3 (or 8) quantisation levels within the uniform scale.
  • the reflection coefficient may be uniformly quantized by observing that the range from -1 to +1 may be evenly divided according to how many bits are available for quantizing a particular reflection coefficient. For example, if there are n bits available to uniformly scalar quantize a reflection coefficient then there may be 2" individual levels to which the reflection coefficient may be quantized.
  • the range from -1 to +1 may be evenly divided such that the interval from -1 to ⁇ 0 may be partitioned into 2" "1 levels.
  • the quantization indices of the quantization levels may range from - 2" "1 to -1.
  • the interval from 0 to +1 may also be evenly partitioned into 2" '1 levels whereby the positive range may be divided such that quantization indices range from 0 to
  • the interval between each quantization level may be the same, in other words a uniform quantization scale. Further, the interval spanned between for any quantization level may be given by the range
  • the quantization level denoted by the quantization index q is the mid point of the quantization interval.
  • each quantization level within the partitioned range from -1 to +1 is given by
  • quantization index q belong to the set of integer numbers [-2" " ',...A..., 2"- 1 -I] .
  • Figure 4 depicts an example of such a uniform quantization scale for the case where n is 4 bits, in other words the uniform quantization scale comprises 16 levels each of equal interval.
  • the quantization level indices for positive values of reflection coefficients are assigned values 0 to 7, which in turn corresponds to quantization level values from 0 to 7/8.
  • quantization level indices for negative values of reflection coefficients are assigned values -1 to -8, which in turn corresponds to quantization level values from -1/8 to -8/8.
  • the quantization interval associated for example with the quantization level index 1 spans the range 0.5/8 to 1.5/8, and that the width of the interval is 1/8.
  • the uniform quantization scale depicted in Figure 4 is a particular example of a uniform quantization scale which may be used to quantize the reflection coefficients according to embodiments of the invention.
  • each reflection coefficient may be allocated a different number of quantization bits, according to the uniform quantization scale used.
  • the process of quantizing a reflection coefficient k takes the form of mapping the received coefficient k to an interval of the uniform quantization scale.
  • the interval to which the reflection coefficient is mapped is denoted by the quantization index q , where the quantization index q is associated with the mid point or quantization level of the mapped quantization interval.
  • the effect of the operation k * 2 n ⁇ l is to map the reflection coefficient k to one of the 2" quantization intervals, and by adding the rounding factor 0.5 and then applying the floor function to the resultant simply ensures that an integer number is returned which reflects the quantization index q of the mapped quantization interval.
  • the maximum positive reflection coefficient value which may be directly represented by this uniform quantization scale are those reflection coefficient values which may be mapped to the largest positive interval.
  • this interval may be envisaged as the quantization interval associated with the final positive quantization level 7/8.
  • all reflection coefficient values falling within the interval (7-0.5)/8 to (7+0.5)/8 may be directly represented by the quantization level 7/8 and consequently the quantization index 7.
  • reflection coefficient values which are near to the value of one are very susceptible to quantization noise or other errors. This is due to the fact that these reflection coefficients are associated with LPC coefficients whose values in the Z-domain lie near to the unit circle, and consequently tend to be associated with large oscillatory regions of the audio signal.
  • the lowest negative interval of the quantization scale may be represented for an n bit quantization scheme by
  • this interval corresponds to the interval from (-8+0.5)/8 to (-8-0.5)/8.
  • the lower region of the most negative quantization interval that is the region from
  • n-l to ⁇ 1 n- ⁇ may be considered to be superfluous since it lies outside the region of the most negative reflection coefficient.
  • this lower region of the most negative quantization interval may be utilised to quantise reflection coefficient values which lie beyond the upper region of the most positive quantization interval.
  • ⁇ n-l to ⁇ 1 n-l may be used to quantize those positive value of reflection coefficients k which lie within the region ( 2 '" - 1 > + ⁇ 5 ⁇ * ⁇ 1 .0
  • positive reflection coefficients in the region (2 " - , n 1 - ) l + ⁇ 5 - ⁇ * ⁇ i.o may be quantized to the quantization level
  • «-i ⁇ fr ⁇ 1.0 may need to be distinguished from the use of the same quantization index to represent negative reflection coefficients in the region
  • This may be effectuated in some embodiments by additionally signalling whether the most negative quantisation index represents either the most negative quantized value, or the most positive quantized value.
  • this additional signalling may not incur extra bits since it may be conveyed inherently as part of further quantization information which is sent as part of the bitstream to the corresponding decoder and hence dequantizer 907 in Figure 6.
  • the reflection coefficient k may be further quantized using a non uniform quantization scheme.
  • This further non uniform quantization scheme may be regarded as an extension to the uniform quantization scheme outlined above.
  • the non uniform quantization extension to the uniform quantization scheme may be applied to each reflection coefficient £ on a scalar, or individual basis. Further, the non uniform quantization scheme may be applied from the overall perspective of reducing the distortion in a predicted residual signal.
  • the predicted residual signal may be determined to be the difference between the predicted signal from the LPC prediction filter and the original signal.
  • the prediction residual signal may also be referred to as the error signal.
  • the prediction residual signal may be viewed as the error signal error signal 234 in Figure 3.
  • the predicted residual signal from a moving average (MA) LPC prediction filter may have a Gaussian distribution, and consequently the distortion measure for this signal may be determined to a squared error distortion.
  • the MA LPC filter may be considered to be the combination of the predictor 220 and the signal subtractor 224 in Figure 3.
  • the squared error distortion may be represented as where i is a sample instance in time, x t is an input sample at instance in time i , and Jc 1 . is the predicted sample instance. Therefore if ⁇ /(*,.,£,.) is the distortion measure per sample, then the distortion between a sequence of n samples X n and the corresponding n predicted sample values X n is the average over the n samples. In other words the distortion measure over the sequence of n samples may be determined to be the mean square error.
  • rate distortion theory the required bit rate required to achieve a particular mean square error distortion may be estimated.
  • the overall bit rate for a given mean square error may be approximated using rate distortion theory as
  • R(d) C + Uog 2 (d(X n ,XJ) where C is a constant.
  • the overall bit rate in terms of the prediction order P of the LPC predictor.
  • the effect of a further reflection coefficient on the residual signal may be determined by considering that the mean square error of a predictor deploying M-1 reflection coefficients will improve by a factor of ⁇ -k 2 , where k is the value of the further reflection coefficient.
  • the overall bit rate required to transmit information at the level of distortion of d ⁇ 1 [X n , X J* (1 - k 2 ) may determined to be
  • R p (d) C +- ⁇ og 2 (d P _ ⁇ (X n ,X n ) * (l -k 2 )) .
  • This value may represent the difference in compression ratio for a value of reflection coefficient k .
  • the difference in compression ratios between successive quantized values of reflection coefficients is a constant. This may be achieved by ensuring that the rate of change of overall bit rate between successive quantized reflection coefficients is also a constant.
  • the rate of change of overall bit rate between successive quantized reflection coefficients may be determined by taking the derivative of the expression for the difference in overall bit rate above.
  • the derivative may be determined to be
  • the inverse of this derivative may be used to refine the quantization step size in the region of the value of the reflection coefficient k .
  • the number of bits used to refine the quantization step size is given by the logarithm of the inverse derivative, in other words the number of bits which maybe used to refine the quantization step size in the region of the reflection coefficient value k may be expressed as
  • B inc may represent a number of fractional bits. For example from the above expression a reflection coefficient value of 0.7071 would result in a B inc value of 0.5, and a reflection coefficient value of 0.9354 would result in a B inc value of 1.5. In such embodiments therefore it may be required to add one refinement bit for reflection coefficient values which exceed 0.7071 , and add a further refinement bit for those reflection coefficient values which exceed 0.9354.
  • the values 0.7071 and 0.9354 may be considered to be thresholds k Ll and k L2 respectively, whereby if exceeded then the uniform quantization interval assigned to that particular reflection coefficient warrants further refinement of the quantization interval step size.
  • the above thresholds may be pre determined and therefore calculated off line from the normal operation of the reflection coefficient quantizer 218. Further since the reflection coefficients exhibit essentially symmetrical values, that is all reflection coefficient values lie within the set of real numbers [-1, 1], the above threshold values may be applied against the absolute value of the reflection coefficient, and therefore apply to negative as well as positive reflection coefficient values.
  • the first stage of the approach is to uniform scalar quantize the reflection coefficient as outlined above, whereby a uniform quantization interval and hence uniform quantization level is assigned to the reflection coefficient.
  • the second stage comprises testing if the reflection coefficient exceeds one of a number of predetermined thresholds. If the reflection coefficient is determined to exceed any one of the thresholds then the step size of the uniform quantization interval assigned to that particular reflection coefficient warrants further refinement.
  • the refinement stage may be considered to be a form of non uniform quantization, since the step sizes of each interval within the quantization scale are no longer the same.
  • the reflection coefficient is first scalar quantized according to some embodiments and then the quantized reflection coefficient is checked against the quantization intervals q L1 and q L2 corresponding to the pre determined thresholds k Ll and k L2 . If it is determined that the quantized reflection coefficient maps to or exceeds a quantization interval designated a threshold interval then the quantization step size of the mapped quantized interval may be further refined with the use of the further refinement bits.
  • a further refinement bit may be used to further divide the interval size of the quantization level associated with the quantization interval index q .
  • the further refinement bit may be used to represent whether the original reflection coefficient k is positioned in the upper half or lower half side of the quantization interval. In other words, if the reflection coefficient is mapped to the quantization interval whose index is q , then the quantization interval may span the range q - OS ⁇ q + OS
  • a value of 0 for the refinement bit may be used to represent that the reflection coefficient k may lie within the lower half of the quantization interval, in other words the sub interval from q -0.5
  • H -i 2 "' 1 ' and a value of 1 for the refinement bit may represent that the reflection coefficient k may lie within the upper half of the quantization interval, in other words the sub interval from
  • the quantized interval index q for the reflection coefficient k may equal or exceed the second threshold q L2 .
  • two further refinement bits may be used to further divide the quantization interval step size into four distinct sub intervals.
  • the four sub intervals of a positive valued reflection coefficient whose quantization interval index may be denoted as q may be expressed as lz ⁇ 5 to l - ⁇ 25
  • the reflection coefficient k may be further quantized by mapping the reflection coefficient value to one of the above sub intervals.
  • This further refined step size may be coded using the two refinement bits. It is to be appreciated in some embodiments that the above described quantization interval refinement method may be equally applied to negative reflection coefficient values as well as positive reflection coefficient values.
  • the number of predetermined thresholds implemented in embodiments of the invention may be greater than the exemplified two threshold system as described above.
  • the above equation for B mc may be used to derive four predetermined threshold values, k Li , k L2 , k LJ and k L4 .
  • the four predetermined threshold values may be derived by setting B mc to 0.5, 1.5, 2.5 and 3.5 which in turn correspond to 1 , 2, 3 and 4 refinement bits respectively.
  • reflection coefficient whose value exceeds k Ll but is less than k L2 , k LJ and k LA may be quantized with a step size interval which is further refined to two sub step intervals, a reflection coefficient whose value exceeds k L2 but is less than k LJ and k L4 may be quantized with an interval which is further refined to four sub step intervals, a reflection coefficient whose value exceeds k L3 but is less than k LA may be quantized with an interval which may be further refined to eight sub intervals, and finally if the reflection coefficient value exceeds the threshold k L4 then the value may be quantized with an interval which may be further refined to 16 sub step intervals.
  • the lower region of the most negative quantization interval may be utilised to quantize reflection coefficient values which lie beyond the upper region of the most positive quantization interval.
  • n-l to > «-l may be used to quantize those positive value of reflection coefficients k which lie within the region
  • the positive reflection coefficients lying within this region may be represented by using the most negative quantisation interval index q - -2" "1 together with the state of the refinement bit associated with the most negative quantisation interval. For instance, if the quantization interval associated with the most negative quantization level - 2" "1 spans the range from then a refinement bit associated with the quantization level may be used to indicate which side of the mid point of the interval the reflection coefficient lies. In such an embodiment the state of the refinement bit which represents the most negative region, in other words the region from
  • the number of uniform quantization intervals within the quantization scale may be determined by the number of bits assigned to uniformly quantize a particular reflection coefficient, and that that the number of uniform quantization intervals may be predetermined in an off line mode of operation before the process of quantizing the reflection coefficients. It is to be further appreciated in some embodiments that the threshold values and those intervals which constitute a threshold interval are also predetermined in an off lime mode of operation before the process of quantizing the reflection coefficient commences.
  • the number of refinement bits assigned to those quantization intervals which are either determined to be a threshold interval or are above a threshold interval may be predetermined, and consequently the partitioning of the intervals into sub regions according to the number of refinement bits may also be predetermined.
  • the step of receiving the set of reflection coefficients from the optimal predictor calculator 216 is depicted as processing step 701 in figure 5.
  • the set of reflection coefficients as determined by the optimal predictor calculator 216 may be delivered to the reflection coefficient quantizer 218 as a set of reflection coefficients, and that the quantizing process is performed on each reflection coefficient on an individual scalar basis.
  • the reflection coefficients may be selected for quantization in prediction order, so reflection coefficients associated with the lower order filter delays may be selected for quantization before those reflection coefficients associated the higher filter order delays.
  • the step of selecting a reflection coefficient for quantizing is shown as processing step 703 in Figure 5.
  • the reflection coefficient may be initially scalar quantized by mapping the coefficient value to the nearest uniform quantization interval as described above.
  • processing step 705 The step of uniform quantizing the reflection coefficient by mapping it to the quantization interval from the quantization scale range is shown as processing step 705 in Figure 5.
  • the determined interval may be checked in order to determine if it is either a threshold interval or above a threshold interval. If the determined quantization interval is equal to a predetermined threshold interval or exceeds a predetermined threshold interval then a number of refinement bits may be allocated to the quantization index. These bits may then be used according to some embodiments to further partition the width of the determined quantization interval in order to produce evenly spaced sub intervals within the determined quantization interval. According to some embodiments the reflection coefficient may then be mapped to the nearest sub interval within the determined quantization interval.
  • the first negative predetermined threshold value has the same absolute value as the first positive predetermined threshold value
  • the second negative predetermined threshold value has the same absolute value as the second positive predetermined threshold value
  • the first negative and first positive predetermined threshold may each have a single allocated refinement bit.
  • the quantisation interval associated with either of these first predetermined thresholds may be further partitioned into two sub regions about the mid point of the quantisation interval.
  • quantization intervals which are higher than the quantization interval associated with the positive predetermined threshold may also be allocated a refinement bit and may accordingly be further partitioned into two sub regions about the mid point of the quantization interval.
  • quantization intervals which are lower than the quantization interval associated with the negative predetermined threshold may also be allocated a refinement bit and accordingly may be further partitioned into two sub regions about the mid point of the quantization interval.
  • the second negative and second positive predetermined thresholds may each be allocated two refinement bits.
  • the quantization interval associated with either of these second predetermined thresholds may be further partitioned into four sub regions by dividing the quantization interval into four equally proportioned sub intervals.
  • quantization intervals which are higher than the quantization interval associated with the second positive predetermined threshold may also be allocated two refinement bits and may therefore be accordingly further partitioned into four sub regions of equal width on the quantization scale.
  • quantization intervals which are lower than the quantization interval associated with the second negative predetermined threshold may also be allocated two refinement bits and accordingly the quantization interval may be further partitioned into four equally proportioned sub intervals.
  • the step of determining if the quantization interval of the uniform quantised reflection coefficient is equal to or above a threshold quantization interval is shown by the determining step 707 in Figure 5.
  • the uniform quantization interval may be further divided into a number of equally proportioned sub intervals according to the number of allocated refinement bits.
  • the reflection coefficient quantizer 218 may then determine if this is the final reflection coefficient to quantize within the set of received reflection coefficients. If it is determined that there are further reflection coefficients for quantization then processing step 703 may be performed again in order to select the next reflection coefficient for quantization. It is to be appreciated in some embodiments that the uniform quantization levels which are equal to or exceed a threshold interval may be divided into sub intervals according to a non linear pattern. In such embodiments the spacing between each sub interval may be apportioned such that the range spanned across each interval may be different.
  • processing step 711 the non even distribution of sub intervals within particular uniform quantisation levels may be pre determined.
  • the step of determining if there are any further reflection coefficients for quantization is depicted as processing step 711 in Figure 5.
  • a quantized reflection coefficient may at least be represented by the index of the quantisation interval.
  • the quantized reflection coefficient may be further represented with an additional at least one refinement bit.
  • these quantized reflection coefficients may be represented with the index of the quantization interval and an additional at least one refinement bit.
  • the output from the reflection coefficient quantizer 218 may comprise the quantization information for each reflection coefficient.
  • the output from the reflection coefficient quantizer 218 may be connected to a quantized reflection coefficient encoder 230 which may in some embodiments encode the quantization information representing the reflection coefficients into a more suitable format for either storage or transmitting as side information.
  • the output from the reflection coefficient quantizer 218 may be connected to the input of the predictor 220.
  • the predictor 220 may comprise a dequantizer which is capable of forming the quantized reflection coefficient value from the received quantization information for each reflection coefficient.
  • the predictor 220 may first for each reflection coefficient decode the uniform quantization value.
  • the information representing the quantized reflection coefficient may further comprise an additional index representing a non uniform component to the uniform quantized coefficient value.
  • the additional index may indicate a particular sub interval within the partitioned uniform quantization interval, and it is the value of the particular sub interval which provides the value of the non uniform quantized reflection coefficient.
  • the method comprises: mapping a coefficient to an interval of an uniform quantization scale; determining whether a value of the coefficient satisfies at least one of at least one p re-determined threshold; and further mapping the coefficient to a non-uniform quantization scale dependant on whether the value of the coefficient satisfies at least one of the at least one pre-determined threshold.
  • This in some embodiments may be therefore be an apparatus comprising: a uniform quantizer configured to map a coefficient to an interval of an uniform quantization scale; an interval threshold determiner configured to determine whether a value of the coefficient satisfies at least one of at least one pre-determined threshold; and a nonuniform quantizer configured to further map the coefficient to a non-uniform quantization scale dependant on the value of the coefficient satisfying at least one of the at least one pre-determined threshold.
  • the additional index representing the non uniform component may be represented by a single bit refinement index.
  • the refinement bit may be used to indicate which of two sub intervals within the equally partitioned quantization interval I k provides the quantized value of the reflection coefficient. In other words if the uniform quantization interval I k spans the range from
  • a first value of the refinement bit may represent a quantized reflection coefficient value given by the mid point of the range from
  • a quantized reflection coefficient value k * associated with the first value of the refinement bit may be determined to be , , / A . - 0.25
  • a second value of the refinement bit may represent a quantized reflection coefficient value given by the mid point of the range from h t0 VtM
  • a quantized reflection coefficient value k * associated with the second value of the refinement bit may be determined to be
  • the additional index representing the non uniform component may be represented by a two bit refinement index.
  • the refinement bits may be used to indicate which of four sub intervals within the equally partitioned quantization interval I k provides the quantized value of the reflection coefficient. In other words if the uniform quantization interval I k spans the range from
  • a first value of the two refinement bits may represent a quantized reflection coefficient value given by the mid point of the range from h -0-5 tQ ⁇ - -0-25
  • a quantised reflection coefficient value k * associated with the first value of the refinement bit may be determined to be
  • the second value of the two refinement bits may represent a quantized reflection coefficient value given by the mid point of the range from
  • a quantized reflection coefficient value k' associated with the second value of the refinement bit may be determined to be
  • the quantized reflection coefficient values may then be converted to their corresponding LPC coefficient values and used as the filter taps for a moving average (MA) predictor of the form
  • *,_,_ ! may represent past audio input samples to the encoder 118, therefore in such embodiments Jc, denotes the predicted current audio sample obtained from a weighted average of past audio input samples.
  • the predicted audio output signal from the predictor 220 may be connected to an input to a compander 222.
  • the compander 222 may be used to apply a non linear compression function to the predicted audio signal, and that the linear compression function applied may have an inverse characteristic to the function applied at the input to the encoder 1104 by the decompander 210.
  • the output from the predictor 220 may be arranged to be connected directly to the input of the subtractor 224. This may particularly be the case when the input to the encoder 104 is an audio signal in the form of linear sampled PCM signal. In such embodiments there may not be a decompander at the input 203 and consequently there may not be a requirement for a compander to be connected to the output of the predictor 220.
  • the output predicted audio signal from the compander 222 may be arranged to be connected to the subtractor 224.
  • the subtractor 224 may be arranged to receive a further input comprising the audio input 203 to the encoder 104.
  • the subtractor 224 may in some embodiments subtract the predicted signal from the audio encoder input signal 203 to provide an output error signal. With reference to Figure 3, the output error signal from the subtractor 224 may then be connected to the input to an entropy coder 226.
  • the entropy coder 226 may encode the error signal by using techniques such as the Huffman coding.
  • the entropy coder 226 may then pass the entropy encoded difference signal to the bitstream formatter 232.
  • bitstream formatter 232 may be arranged to receive the encoded indices and refinement bits from the encoder quantized reflection coefficient encoder 230. The bitstream formatter 232 may then be further arranged to format the received bitstreams to produce the bitstream output 112 from the encoder 104. In some embodiments of the invention the bitstream formatter 232 may interleave the received inputs and may generate error detecting and error correcting codes to be inserted into the bitstream output 112.
  • an apparatus comprising at least one processor and at least one memory including computer program code the at least one memory and the computer program code configured to, with the at least one processor, cause the apparatus at least to perform: mapping a coefficient to an interval of an uniform quantization scale; determining whether a value of the coefficient satisfies at least one of at least one pre-determined threshold; and further mapping the coefficient to a non-uniform quantization scale dependant on whether the value of the coefficient satisfies at least one of the at least one pre-determined threshold.
  • the decoder 108 receives the encoded signal stream 112 comprising the encoded information representing the quantised reflection coefficients and the entropy encoded difference signal and outputs a reconstructed audio signal 114.
  • the decoder comprises an input 902 by which the encoded bitstream 112 may be received.
  • the input 902 may be connected to a bitstream unpacker or de-multiplexer 901 which may receive the encoded signal and output the entropy encoded error signal and the encoded information representing the quantized reflection coefficients.
  • the entropy encoded error signal from the bitstream unpacker 901 may be connected to an input to an entropy decoder 903.
  • the entropy decoder 903 may produce the error signal which may then be connected to an input to an adder 905.
  • the encoded information representing the quantized reflection coefficients from the bitstream unpacker 901 may be connected to a quantized reflection coefficient decoder 907.
  • the quantized reflection coefficient decoder 907 in some embodiments may decode the received signal in order to produce the information representing the quantised reflection coefficients.
  • the information representing the quantized reflection coefficients may comprise at least the quantization index representing the uniform quantization component for each reflection coefficient.
  • information representing the quantized reflection coefficients may further comprise for some of the reflection coefficients a further non uniform quantization component represented by the refinement bits.
  • the information representing the quantized reflection coefficients may then be passed to the reflection coefficient dequantizer 909.
  • the dequantizer 909 may first for each reflection coefficient decode the uniform quantization value.
  • the information representing the quantized reflection coefficient may further comprise an additional refinement index representing a non uniform component to the uniform quantized coefficient value.
  • the additional refinement index may indicate a particular sub interval of the uniform quantization interval. It is to be understood that it is the value associated with the particular sub interval which provides the value of the non uniform quantized reflection coefficient.
  • the non uniform quantized reflection coefficient value may be determined as the value associated with the mid point of the particular sub interval. It is to be understood that the process of determining the value of the quantized reflection coefficient may be performed for each reflection coefficient in turn, and that the quantized value of the reflection coefficient will either be an interval value according to an uniform quantization scale, or a value associated with a sub interval within an interval of the uniform quantization scale.
  • the predictor 911 may first convert the set of received quantized reflection coefficients into the corresponding quantized LPC coefficients.
  • the set of quantized reflection coefficients received by the predictor 911 will have a number of elements equivalent to the order of the subsequent prediction filter.
  • the set of quantized reflection coefficients will correspond with a frame of output audio samples.
  • the quantized reflection coefficients may then be used as the filter taps in the following LPC prediction filter
  • ⁇ * corresponds to the quantized LPC coefficient
  • X 1- ⁇ 1 represents past reconstructed sample values
  • P denotes the order of the filter and therefore there are P LPC coefficients in total
  • X 1 represents the predicted sample obtained from the weighted sum of the past reconstructed samples.
  • the output from the predictor may be the predicted sample value, this output signal may be connected to a further input to the adder 905.
  • the adder 905 may then add the predicted sample value to the decoded error sample value from the entropy decoder 903.
  • the output from the adder may then form the output audio signal 1 14 from the decoder 108.
  • USB universal serial bus
  • modem data cards may comprise audio enhancement apparatus such as the apparatus described in embodiments above.
  • user equipment is intended to cover any suitable type of wireless user equipment, such as mobile telephones, portable data processing devices or portable web browsers.
  • PLMN public land mobile network
  • the various embodiments described above may be implemented in hardware or special purpose circuits, software, logic or any combination thereof.
  • some aspects may be implemented in hardware, while other aspects may be implemented in firmware or software which may be executed by a controller, microprocessor or other computing device, although the invention is not limited thereto.
  • firmware or software which may be executed by a controller, microprocessor or other computing device, although the invention is not limited thereto.
  • While various aspects of the invention may be illustrated and described as block diagrams, flow charts, or using some other pictorial representation, it is well understood that these blocks, apparatus, systems, techniques or methods described herein may be implemented in, as non-limiting examples, hardware, software, firmware, special purpose circuits or logic, general purpose hardware or controller or other computing devices, or some combination thereof.
  • the embodiments of the application may be implemented by computer software executable by a data processor, such as in the processor entity, or by hardware, or by a combination of software and hardware.
  • a data processor such as in the processor entity, or by hardware, or by a combination of software and hardware.
  • any blocks of the logic flow as in the Figures may represent program steps, or interconnected logic circuits, blocks and functions, or a combination of program steps and logic circuits, blocks and functions.
  • the software may be stored on such physical media as memory chips, or memory blocks implemented within the processor, magnetic media such as hard disk or floppy disks, and optical media such as for example digital versatile disc (DVD), compact discs (CD) and the data variants thereof both.
  • DVD digital versatile disc
  • CD compact discs
  • the memory may be of any type suitable to the local technical environment and may be implemented using any suitable data storage technology, such as semiconductor-based memory devices, magnetic memory devices and systems, optical memory devices and systems, fixed memory and removable memory.
  • the data processors may be of any type suitable to the local technical environment, and may include one or more of general purpose computers, special purpose computers, microprocessors, digital signal processors (DSPs), application specific integrated circuits (ASIC), gate level circuits and processors based on multi-core processor architecture, as non-limiting examples.
  • Embodiments of the inventions may be practiced in various components such as integrated circuit modules.
  • the design of integrated circuits is by and large a highly automated process. Complex and powerful software tools are available for converting a logic level design into a semiconductor circuit design ready to be etched and formed on a semiconductor substrate.
  • circuitry may refer to all of the following: (a) hardware-only circuit implementations (such as implementations in only analogue and/or digital circuitry) and (b) to combinations of circuits and software
  • processor(s) any combination of processor(s) or (ii) to portions of processor(s)/software (including digital signal processor(s)), software, and memory(ies) that work together to cause an apparatus, such as a mobile phone or server, to perform various functions) and
  • circuits such as a microprocessor(s) or a portion of a microprocessor(s), that require software or firmware for operation, even if the software or firmware is not physically present.
  • circuitry would also cover an implementation of merely a processor (or multiple processors) or portion of a processor and its (or their) accompanying software and/or firmware.
  • circuitry would also cover, for example and if applicable to the particular claim element, a baseband integrated circuit or applications processor integrated circuit for a mobile phone or a similar integrated circuit in server, a cellular network device, or other network device.
  • processor and memory may comprise but are not limited to in this application: (1 ) one or more microprocessors, (2) one or more processor(s) with accompanying digital signal processor(s), (3) one or more processor(s) without accompanying digital signal processor(s), (3) one or more special-purpose computer chips, (4) one or more field-programmable gate arrays (FPGAS), (5) one or more controllers, (6) one or more application-specific integrated circuits (ASICS), or detector(s), processor(s) (including dual-core and multiple-core processors), digital signal processor(s), controller(s), receiver, transmitter, encoder, decoder, memory (and memories), software, firmware, RAM, ROM, display, user interface, display circuitry, user interface circuitry, user interface software, display software, circuit(s), antenna, antenna circuitry, and circuitry.
  • FPGAS field-programmable gate arrays
  • ASICS application-specific integrated circuits

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Spectroscopy & Molecular Physics (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)
  • Compression, Expansion, Code Conversion, And Decoders (AREA)

Abstract

L'invention concerne un dispositif comprenant un processeur configuré pour mapper un coefficient sur un intervalle d'une échelle de quantification uniforme ; déterminer si une valeur du coefficient satisfait au moins un d'au moins un seuil prédéterminé ; et mapper ensuite le coefficient sur une échelle de quantification non-uniforme en tenant compte du fait que la valeur du coefficient satisfait ou non au moins un de l'au moins un seuil prédéterminé.
PCT/EP2009/058438 2009-07-03 2009-07-03 Dispositif WO2011000434A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/EP2009/058438 WO2011000434A1 (fr) 2009-07-03 2009-07-03 Dispositif

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/EP2009/058438 WO2011000434A1 (fr) 2009-07-03 2009-07-03 Dispositif

Publications (1)

Publication Number Publication Date
WO2011000434A1 true WO2011000434A1 (fr) 2011-01-06

Family

ID=41609784

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2009/058438 WO2011000434A1 (fr) 2009-07-03 2009-07-03 Dispositif

Country Status (1)

Country Link
WO (1) WO2011000434A1 (fr)

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6246979B1 (en) * 1997-07-10 2001-06-12 Grundig Ag Method for voice signal coding and/or decoding by means of a long term prediction and a multipulse excitation signal

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6246979B1 (en) * 1997-07-10 2001-06-12 Grundig Ag Method for voice signal coding and/or decoding by means of a long term prediction and a multipulse excitation signal

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
GHIDO F: "Optimal Quantized Linear Prediction Coefficients for Lossless Audio Compression - Scalar Quantization Revisited", PROCEEDINGS AES 120TH, 22 May 2006 (2006-05-22), pages 1 - 9, XP002567591, Retrieved from the Internet <URL:http://www.aes.org/e-lib/inst/download.cfm/13561.pdf?ID=13561> [retrieved on 20100209] *
LIEBCHEN T: "An introduction to MPEG-4 audio lossless coding", ACOUSTICS, SPEECH, AND SIGNAL PROCESSING, 2004. PROCEEDINGS. (ICASSP ' 04). IEEE INTERNATIONAL CONFERENCE ON MONTREAL, QUEBEC, CANADA 17-21 MAY 2004, PISCATAWAY, NJ, USA,IEEE, PISCATAWAY, NJ, USA, vol. 3, 17 May 2004 (2004-05-17), pages 1012 - 1015, XP010718364, ISBN: 978-0-7803-8484-2 *
UN C K ET AL: "PIECEWISE LINEAR QUANTIZATION OF LPC REFLECTION COEFFICIENTS.", ACOUSTICS, SPEECH, AND SIGNAL PROCESSING, IEEE INTERNATIONAL CONFERENCE ON ICASSP '77, vol. 2, 1977, pages 417 - 420, XP002567306 *
VISWANATHAN R ET AL: "Quantization properties of transmission parameters in linear predictive systems", IEEE TRANSACTIONS ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING USA, vol. ASSP-23, no. 3, 1 June 1975 (1975-06-01), pages 309 - 321, XP002567307, ISSN: 0096-3518 *

Similar Documents

Publication Publication Date Title
RU2608878C1 (ru) Регулировка уровня во временной области для декодирования или кодирования аудиосигналов
KR101278805B1 (ko) 엔트로피 코딩 방법 및 엔트로피 디코딩 방법
US7684981B2 (en) Prediction of spectral coefficients in waveform coding and decoding
US10121480B2 (en) Method and apparatus for encoding audio data
US7693709B2 (en) Reordering coefficients for waveform coding or decoding
RU2329549C2 (ru) Устройство и способ определения величины шага квантователя
CA2651745C (fr) Codage de signaux d&#39;information
US4704730A (en) Multi-state speech encoder and decoder
WO2008021247A9 (fr) Mise en forme arbitraire d&#39;une enveloppe de bruit temporelle sans information secondaire
EP2915166B1 (fr) Procédé et appareil pour quantification vectorielle résiliente
US8665945B2 (en) Encoding method, decoding method, encoding device, decoding device, program, and recording medium
US9548056B2 (en) Signal adaptive FIR/IIR predictors for minimizing entropy
WO2014013294A1 (fr) Codeur de signal audio stéréo
US20100250260A1 (en) Encoder
US20130346073A1 (en) Audio encoder/decoder apparatus
WO2011000434A1 (fr) Dispositif
Paraskevas et al. A Prony method for high-quality audio coding
Sung Finite Wordlength Effects Evaluation of the MPEG-2 Audio Decoder

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 09780148

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 09780148

Country of ref document: EP

Kind code of ref document: A1