WO2008086920A1 - Disturbance reduction in digital signal processing - Google Patents

Disturbance reduction in digital signal processing Download PDF

Info

Publication number
WO2008086920A1
WO2008086920A1 PCT/EP2007/063598 EP2007063598W WO2008086920A1 WO 2008086920 A1 WO2008086920 A1 WO 2008086920A1 EP 2007063598 W EP2007063598 W EP 2007063598W WO 2008086920 A1 WO2008086920 A1 WO 2008086920A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
lpc
perturbation
speech
coefficients
Prior art date
Application number
PCT/EP2007/063598
Other languages
French (fr)
Inventor
Christophe Beaugeant
Herve Taddei
Emmanuel Rossignol Thepie Fapi
Original Assignee
Nokia Siemens Networks Gmbh & Co. Kg
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 Siemens Networks Gmbh & Co. Kg filed Critical Nokia Siemens Networks Gmbh & Co. Kg
Publication of WO2008086920A1 publication Critical patent/WO2008086920A1/en

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
    • G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
    • G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
    • G10L21/0208—Noise filtering
    • 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

Definitions

  • the invention relates to disturbance reduction in digital signal processing.
  • the hands-free systems present the worse case considering noise and echo problems. Indeed, in this case the microphone is far from the talker so that the speech signal is less energetic and the perturbation gains importance. Moreover, the loudspeaker signal is also louder than in non-hands-free use cases so that the coupling between the two transducers (microphone/loudspeaker) increases .
  • Digital telecommunication systems include speech coding. Speech codecs are definitely perturbed by the presence of noise and echo. Indeed, they are optimized to handle single speech signals.
  • LPC Linear Prediction Coefficients
  • P stands for the prediction order. After filtering the input signal by the LPC filter, a residual signal is obtained. This signal needs to be transmitted for reconstruction of the original signal to the decoder.
  • Noise reduction and echo cancellation are historically built as pre-processing before coding the speech.
  • Many solutions reducing the perturbations on PCM signals are available.
  • a state of the art overview can be found in "Combined Noise and Echo Reduction in Hands-Free systems: A Survey” by R. Le Bouquin Jeannes, P. Scalart, G. Faucon, C. Beaugeant; IEEE Trans. On Speech and Audio Processing; vol.9; Nov 2001; pp 808-820.
  • Such solutions are efficient when the PCM data is available, so typically if the problem are solved within the terminal itself, before the encoding of the signal.
  • LPC Linear Prediction Coefficients
  • the digital signal y(n) comprises a useful signal s (n) and a perturbation signal p (n) .
  • the perturbation signal p (n) derives e.g. from noise or echo and includes everything of y(n) that is not part of the useful signal s (n) .
  • the bitstream y e (n) is derived from y(n) by LPC-encoding.
  • LPC Linear Prediction Coefficients
  • Other parameters of the bitstream y e (n) may also be received, like the fixed gain or the adaptive gain of the bitstream y e (n) .
  • the complete bitstream y e (n) is received.
  • the autocorrelation matrix T s of the useful signal s (n) , of the autocorrelation matrix T p , of the perturbation signal p (n) and the LPC A p of the perturbation signal p (n) are estimated.
  • a modified LPC A s is calculated. It is calculated from A and the estimated F 5 , Y pl A p .
  • a modified data stream y e ' (n) including the modified LPC A s is output.
  • This data stream can be received by a decoder which decodes the original signal y(n).
  • Codecs for transmission of speech are optimized for speech signals.
  • the addition of noise or of echo to the useful speech signal leads to sub-optimal behaviour of the codecs, which means additive artefacts on the decoded signal and lower quality.
  • the use of LPC coefficients that are influenced by the noise signal makes the quality of the received speech worse. Accordingly, noise and echo are not only adding undesired information to the useful signal, they also lead to sub-optimal behaviour of speech codecs, decreasing all the quality of telecommunication.
  • a s is preferably calculated by A s .
  • the residual signal is the signal that is obtained after the LPC filtering.
  • y e (n) comprises of the residual signal and the LPC coefficients.
  • the estimations of F s , Y p , A p can be done by classical methods, e.g. by frequency analysis of the encoded signal y e (n) .
  • the method also comprises a step of a noise reduction on the residual signal of the encoded signal y e (n) .
  • a noise reduction technique on residual signals is described in the above-mentioned "Compressed Domain Noise Reduction and Echo Suppression for Network Speech" .
  • the invention described here provides a solution to achieve a reduction of perturbation, like noise and echo, by modifying the LPC coefficients computed during LPC analysis.
  • the Linear Prediction Cofficients (LPC) A y of the signal y e (n) are not received, but calculated from the digital sample signal y(n) .
  • the encoding and modifying the LPC coefficients is done only once. Therefore, the residual signal does not need to be encoded and output twice. This improves the speed for encoding and modifying the LPC.
  • the invention may be used with any system based on model of Eq (1) where additive perturbation disturbed the coefficients
  • the method is applicable in a broad range of applications in signal processing.
  • One possible application where the LPC modification would be useful is earthquake detection.
  • the method is especially qualified for signal transmission in telecommunication. Because the different signal characteristics of voice and noise signals, the autocorrelation matrix of the perturbation signal can be estimated relatively precisely. This ensures that the cleaning of the LPC parameters is made successful.
  • the invention also relates to a digital signal transmission apparatus that performs the inventive method.
  • Such an apparatus comprises means for receiving the LPC coefficients A y , for the estimation of Y s , Y p , A p , for the calculation and output of A p .
  • DSP digital signal processor
  • a corrupted signal y ⁇ n is the sum of a useful signal s(n) with a perturbation p(n) .
  • ⁇ s(n)s(n-j) - ⁇ a s (k)-s(n-k)s(n-j)
  • T T ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇
  • the two last expressions show that the computation of the useful signal LPC is obtained when the following entities are known: the LPC of the perturbed signal (A y ), the LPC of the perturbation (A p ), the covariance matrix of the perturbation (T p ), and the inverse of the covariance matrix of the useful signal ( T s ) .
  • the present invention proposes a method based on formula Eq. (20) -(21) or on any formula derived from this equation to obtain the LPC coefficients of the useful signal (A 5 ), when the LPC coefficients of the perturbed signal (A ) are available.
  • Eq. (20) /(21) require to know the LPC coefficients of the perturbation A p , the correlation matrix of the perturbation T p and the inverse of the correlation
  • the LPC A are available for each frame m.
  • the estimation can be based on the additive information. For instance, when placing the method within a speech codec, other speech codec parameters can be used to obtain the estimation.
  • the estimation can be based on the additive information. For instance, when placing the method within a speech codec, other speech codec parameters can be used to obtain the estimation.
  • the estimation can be based on the additive information. For instance, when placing the method within a speech codec, other speech codec parameters can be used to obtain the estimation. Applying the filter defined in Eq (20) /(21) to get A s .
  • This process can be applied on speech codec bitstream by applying the following steps: - For each frame m, extracting the LPC coefficients A from the speech codec bitstream.
  • a typical application of the method is processing noise reduction, echo reduction or reduction of any other perturbation on the LPC on speech.
  • the methods permits to re-construct the useful LPC parameters without the need to decode the bitstream to get the PCM data and apply classical noise reduction or echo cancellation. This is an alternative solution to existing prior art solutions.
  • Equation (21) The mathematical expression obtained in Equation (21) is relatively easy to implement. It requires of course estimations of certain entities like cross-correlation functions or LPC or the noisy signal, but such estimations are quite classical. The method is accordingly quite classical from a signal processing point of view, and possible to implement in real time applications.
  • Figure 1 shows a signal transmission from a sender to a receiver in a telecommunication system.
  • Figure 2 is a flow chart for the modification of LPC coefficients according to a first embodiment.
  • Figure 3 shows a comparison of transfer function with non- modified LPC versus modified LPC coefficients.
  • Figure 4 shows a second embodiment for the modification of LPC coefficients.
  • Figure 1 shows an embodiment of a telecommunication system 1 in a signal transmission with modified LPC coefficients.
  • the sender 2 generates the useful signal s (t) by talking.
  • Perturbations generate a perturbation signal p(t) with is added to the useful signal resulting in the signal y(t) .
  • the signal y(t) is digitalized in the Analog-Digital-Converter (AD-Converter) 3 which generates a digital signal y(n) .
  • the digital signal y(n) is encoded in the encoder to the signal y e (n) .
  • the encoding is done with the help of an LPC analysis.
  • the encoded signal y e (n) is transmitted via the transmission block 5 to the decoder 6.
  • the decoder 6 receives the signal y e ' (n) from the transmission block and decodes y e ' (n) to a digital signal y d (n).
  • y e ' (n) is either equal or unequal to y e (n) .
  • the transmission block 5 is e.g. a telephone switch, a router or a simple wire.
  • y d (n) is finally DA-converted by the DA-converter to y a (t) which is received as an analog signal by the receiver 8.
  • the modification of the LPC parameters is done in the transmission block 5, whereas in the embodiment of Figure 4, the encoder 4 directly modifies the LPC parameters .
  • Figure 2 is a flow chart for the modification of LPC coefficients within the transmission block 5.
  • y e (n) is a bitstream including LPC coefficients. If the encoder uses the AMR codec, the LPC coefficients are transmitted as Line
  • the frames of y e (n) also comprise the parameters pitch delay, fixed codebook index, fixed gain and adaptive gain.
  • the bitstream is computed by the analysis of successive frames, each, each comprising a defined number of samples (thoughy 160) . If the signal y e (t) is sampled at a frequency of 8 MHz, in the so-called narrow band, the number of LPC coefficients is chosen to 8 or 10 in current standardized codecs (AMR, EFR, FR) . In other words, the codec uses 8th respectively a 10th order linear prediction filter. In Eq. (1) k runs from 1 to 9 respectively from 1 to 11.
  • the sampling frequency is 16 kHz and the number of coefficients is preferably chosen to 16 in current standardized codecs (AMR- WB) .
  • the Figure 2 shows the flow chart where the LPC coefficients are extracted from the bitstream y e (n) .
  • the bitstream is divided in the LPC coefficients and the rest of the bitstream, including the information needed to decode the residual waveform.
  • the estimations of A p , T p and T s ⁇ are applied, taking into account the LPC coefficient as well as eventually additive information from the bitstream.
  • a p is generated by the help of a Voice
  • VAD Voice Activity Detection
  • a p (m) is calculated by the following algorithm, whereby m is the index of the frame or subframe.
  • the perturbation is assumed to be white noise. Accordingly, the autocorrelation matrix T p has the following form:
  • E (m) is the energy of the signal y(n)
  • m indicates a frame or subframe
  • ⁇ is a fixed parameter being heuristically chosen, 0 ⁇ /? ⁇ l.
  • T y may be calculated with the help of Eq. 15 if the data stream is decoded.
  • the bitstream of the encoded signal has to be decoded to make the estimation of T p , T s .
  • this matrices and vectors can also be done on the basis of the codec parameters of the signal y e (n) by interpreting the fixed gain and the adaptive gain.
  • the clean LPC coefficients A s are generated by one of the equations 21 or 22. It should be noticed that the calculated clean LCP A s are an estimation of a LPC of the useful signal s (n) . Accordingly, the calculated LPC A s are as good as the estimations for
  • the filter on the LPC A coefficients is applied to get the clean LPC A s and finally the LPC are replaced in the bitstream by changing each frame by the use of the clean LPC parameters A s .
  • the frames are modified sequentially and sent to the decoder as signals y e ' (n) .
  • This method of improving the speech signal quality can be done anywhere in the path between the encoder and decoder.
  • the method can be applied in the terminal of the sender, in the terminal of the receiver or in one of the routers telephone switches or gateways between different networks.
  • the use of the modified LPC coefficients improves the quality of the received signal which is demonstrated with the help of Figure 3.
  • Figure 3 shows a comparison of transfer functions with non- modified LPC versus modified LPC coefficients.
  • the synthesis LPC filter function can be described by the filter transfer function H(f) in the frequency domain.
  • the graph of Fig. 3 shows a functions H(f) dependent on the frequency f for a non-noisy LPC function and, with the dashed line, for a noisy LPC filter.
  • the transfer function of the noisy LPC filter in our case a non-modified LPC, has more energy but is smoother.
  • Using a LPC that was generated on the basis of a noisy signal worsens the quality of the speech.
  • the modification of the LPC to a clean LPC make it easier for the receiver to understand the received speech and enhance the clarity of the speech.
  • Figure 2 may be extended by an additional step which reduces noise on the rest bitstream. This noise reduction is performed after the estimation of A p , T p and T ⁇ and before the generation the new frames of the bitstream of the signal.
  • Figure 4 shows a second embodiment of the modification of LPC coeffients. In this case, the function is included in the encoder 4.
  • the LPC coefficient are computed by an analysis of successive weighted frames.
  • the Levinson-Durbin algorithm permits to get the LPC coefficient from the sample y(n) of the analysis frame.
  • Our method maybe placed as a postfilter of the computation blocks of the LPC analysis. In this scenario, it enhances the LPC coefficient by reducing the influence of the noise.
  • the needed estimations (A ⁇ , T p and
  • T s may be done by using the LPC coefficients but also some additive information of sample y(n) as depicted in Fig 4. Finally, the filter of Eq (20) /(21) is applied on the perturbed coefficients to get the enhanced ones.
  • the method of improving the speech quality is performed within the encoder.
  • the encoder receives the samples from the A/D-converter .
  • the samples are organized as frames.
  • the LPC analysis outputs the LCP coefficients A y .
  • T p and T s ⁇ are estimated like one of the embodiments described above.
  • LPC coefficients A s are calculated by one of the equations (21) and (22) .
  • the encoding of the frame is done with A .
  • Reference number list

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)
  • Quality & Reliability (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Compression, Expansion, Code Conversion, And Decoders (AREA)

Abstract

A method is provided for transmitting a digital signal y(n), y(n) comprising a useful signal and a perturbation signal p(n). The method comprises the steps of: receiving the Linear Prediction Coefficients (LPC) Ay of the a signal ye(n), ye(n) being an LPC-encoded signal of y(n); estimating the autocorrelation matrix rs of the useful signal s(n), of the autocorrelation matrix rp of the perturbation signal p(n) and the LPC Ap of the perturbation signal p(n); calculating a modified LPC As by using Ay and the estimated rs, rp, Ap; outputting a modified data stream ye'(n) including the modified LPC As.

Description

Description
Disturbance reduction in digital signal processing
The invention relates to disturbance reduction in digital signal processing.
In telecommunication devices, sound pick-up has to deal with two problems: The presence of noise in the environment where the devices are and the echo phenomenon due to the coupling between the loudspeaker and the microphone. These phenomenon decrease the quality of the communication: On one hand, the speech signal of the user is corrupted by the environmental noise which leads to tiredness for the far-end speaker and in case of too loud noise to misunderstanding between the correspondents. The echo phenomenon is also really perturbing for the far-end speaker as he can hear his own voice due to the coupling. If the transmission delay of the communication network is important (more than 30 ms) , he has the really bad feeling of hearing his own voice delayed. Current mobile networks and their delay of more than 150 ms make this echo phenomenon annoying (previously, fixed telephony network provided low delay for local calls and the echo phenomenon was disturbing only for international communications) .
These two drawbacks are inherent to phone communication. In case of handset or headset, their effects on the far-end speaker are a little bit smoothed by the fact that the speech of the user is generally more energetic than the two perturbations (noise and echo) . Nevertheless, in certain conditions (in stations, on the street) , the surrounding noise can really disturb the far-end speaker. The echo can also appear in small recent devices, or in headsets. In both cases the coupling exists and the echo can be energetic enough to be disturbing.
The hands-free systems present the worse case considering noise and echo problems. Indeed, in this case the microphone is far from the talker so that the speech signal is less energetic and the perturbation gains importance. Moreover, the loudspeaker signal is also louder than in non-hands-free use cases so that the coupling between the two transducers (microphone/loudspeaker) increases .
Digital telecommunication systems include speech coding. Speech codecs are definitely perturbed by the presence of noise and echo. Indeed, they are optimized to handle single speech signals.
There exists a lot of codecs, and many of them are using Linear Prediction Coefficients (LPC) analysis. For example, CELP codecs as AMR, EFR, MELP, TCX or vocoders use such an analysis. The generic principle of the LPC analysis is to provide a linear estimation of the input speech signal y(n) through an AR filter as follow:
y(n) = -∑ay(k).y(n-k) (I] k=\
Where [#.,,(&)] stand for the Linear Prediction Coefficients and
P stands for the prediction order. After filtering the input signal by the LPC filter, a residual signal is obtained. This signal needs to be transmitted for reconstruction of the original signal to the decoder.
Noise reduction and echo cancellation are historically built as pre-processing before coding the speech. Many solutions reducing the perturbations on PCM signals are available. A state of the art overview can be found in "Combined Noise and Echo Reduction in Hands-Free systems: A Survey" by R. Le Bouquin Jeannes, P. Scalart, G. Faucon, C. Beaugeant; IEEE Trans. On Speech and Audio Processing; vol.9; Nov 2001; pp 808-820. Such solutions are efficient when the PCM data is available, so typically if the problem are solved within the terminal itself, before the encoding of the signal.
Within the communication chain, it may happen that only the coded signal is available. This is the case in any part of the network where only the codec bitstream is available. It may also be the case in terminal where an integrated chipset encodes the microphone signal and only delivers a bitstream to the processor where noise reduction or echo cancellation can be implemented.
In such scenarios, classical solutions can only be applied by decoding the signal, processing it and re-encoding. This leads to a suboptimal solution compared to any solution that would be based on the processing of the PCM signal. Indeed, decoding and re-encoding lead to high computation load, moreover, the so-called tandem effect due to decoding-re- encoding decreases the quality of the speech signal. Accordingly, the enhancement obtained by noise reduction and echo cancellation may be compensated by the artefacts introduced by the tandem effect.
More recently, the idea appeared to modify the codec parameter "fixed gain" in A-CELP coding to decrease of the signal energy when perturbation is detected. This is shown in - "Compressed Domain Noise Reduction and Echo Suppression for Network Speech" by Chandran, Ravi and Marchok, Daniel J.; Enhancement, Proc.of the 43rd IEEE Midwest Symposium on Circuits and Systems, pp 10-13, 2000,
- "Noise reduction on speech codec parameters" by Herve Taddei, Christophe Beaugeant, Michael de Meuleneire; ICASSP 2004
- and "Gain Loss Control based on Speech Codec Parameters" by C. Beaugeant, N. Duetsch, H. Taddei; Eusipco 2004.
Such techniques allow an efficient decrease of the perturbation energy but can not reduce all the artefacts present on the speech.
It is an object of the invention to provide a method for reducing noise on encoded digital signals being encoded with Linear Prediction Coefficients (LPC) .
This object is solved by the subject-matter of the independent claims. Further enhancements are provided by the subject-matter of the dependent claims.
A method is provided for transmission of a digital signal y(n). The digital signal y(n) comprises a useful signal s (n) and a perturbation signal p (n) . The perturbation signal p (n) derives e.g. from noise or echo and includes everything of y(n) that is not part of the useful signal s (n) .
The bitstream ye (n) is derived from y(n) by LPC-encoding. As a first step, the Linear Prediction Coefficients (LPC) A of a bitstream ye (n) are received. Other parameters of the bitstream ye (n) may also be received, like the fixed gain or the adaptive gain of the bitstream ye (n) . As an option, the complete bitstream ye (n) is received. The autocorrelation matrix Ts of the useful signal s (n) , of the autocorrelation matrix Tp, of the perturbation signal p (n) and the LPC Ap of the perturbation signal p (n) are estimated.
A modified LPC As is calculated. It is calculated from A and the estimated F5, Ypl Ap . In an output step, a modified data stream ye ' (n) including the modified LPC As is output.
This data stream can be received by a decoder which decodes the original signal y(n).
Codecs for transmission of speech are optimized for speech signals. The addition of noise or of echo to the useful speech signal leads to sub-optimal behaviour of the codecs, which means additive artefacts on the decoded signal and lower quality. The use of LPC coefficients that are influenced by the noise signal makes the quality of the received speech worse. Accordingly, noise and echo are not only adding undesired information to the useful signal, they also lead to sub-optimal behaviour of speech codecs, decreasing all the quality of telecommunication.
Our solution is based on this innovative principle to modify directly the parameter computed by the speech encoders. Compared to prior art solutions, it proposes to modify the LPC coefficients. The received speech sounds more precise if the LPC coefficients are less influenced by noise.
As is preferably calculated by As
Figure imgf000006_0001
.
This equation only comprises multiplication and addition functions and is accordingly adapted for digital signal processors. As also can be calculated by the equivalent equation As =Ay +T~lTp[Ay -Ap] . If also the residual signal of the encoded signal ye (n) is received, the steps of estimating Fs, Yp, Ap can be based on the residual signal of ye (n) and A . The residual signal is the signal that is obtained after the LPC filtering. ye (n) comprises of the residual signal and the LPC coefficients. The estimations of Fs, Yp, Ap can be done by classical methods, e.g. by frequency analysis of the encoded signal ye (n) .
In an embodiment, the method also comprises a step of a noise reduction on the residual signal of the encoded signal ye (n) . This gives the additional advantage that the noise is reduced in the output data stream ye ' (n) . A noise reduction technique on residual signals is described in the above-mentioned "Compressed Domain Noise Reduction and Echo Suppression for Network Speech" . The invention described here provides a solution to achieve a reduction of perturbation, like noise and echo, by modifying the LPC coefficients computed during LPC analysis.
In another embodiment, the Linear Prediction Cofficients (LPC) Ay of the signal ye (n) are not received, but calculated from the digital sample signal y(n) . The encoding and modifying the LPC coefficients is done only once. Therefore, the residual signal does not need to be encoded and output twice. This improves the speed for encoding and modifying the LPC.
Even if the embodiments of the invention will be described in respect to speech signals, the invention may be used with any system based on model of Eq (1) where additive perturbation disturbed the coefficients
Figure imgf000007_0001
The method is applicable in a broad range of applications in signal processing. One possible application where the LPC modification would be useful is earthquake detection.
But, the method is especially qualified for signal transmission in telecommunication. Because the different signal characteristics of voice and noise signals, the autocorrelation matrix of the perturbation signal can be estimated relatively precisely. This ensures that the cleaning of the LPC parameters is made successful.
The invention also relates to a digital signal transmission apparatus that performs the inventive method. Such an apparatus comprises means for receiving the LPC coefficients Ay , for the estimation of Ys, Yp, Ap , for the calculation and output of Ap .
Especially, the calculation of As is preferable done by a digital signal processor (DSP) because a DSP is effective in performing multiplications and additions.
The invention deals with a method to reconstruct the LPC coefficient Ai=[αi(A:)]r of the useful signal s(n) knowing the LPC coefficients A
Figure imgf000008_0001
computed on a perturbed signal y{n) . It is based on the following mathematical development of the LPC analysis.
We consider that a corrupted signal y{n) is the sum of a useful signal s(n) with a perturbation p(n) . This perturbation can be additive noise, echo or more generally any signal that is not desired: y(n) = s(n) + p(n) (2)
We assume that the perturbation and the useful signal are not correlated.
We also assume that an LPC analysis of order P is applied on analysis frames of TV samples (this is the case for speech codecs based on LPC) , so that the signal y(n) is estimated by y{n) as :
P y(n) =-∑ay(k).y(n-k) (3] k=\
In the same way, the useful signal s(n) is modeled as:
P s(n) =-∑as(k).s(n-k) (4) k=\ and, an estimation pin) ofp(n) can be written with:
p(n) = -∑ap(k).p(n-k) (5) k=\
Let's consider the following squared error
JV-I
Esr = ∑ y(n)+∑ay(k)-y(n-k) :6)
B=O k=\
Using the fact that the signal y(n) can be written as the sum of the useful signal and of the perturbation, we can write:
:
Figure imgf000009_0001
The error Esτ is minimum with respect to the LPC coefficients A when its derivative with respect to each coefficient is zero :
dE
V/, ST _ = 0 0 daJj)
It leads to
Figure imgf000010_0001
(9
This can be written equivalently
JV-I
∑[sin)sin-j)+pin)pin-J)+sin)pin-J)+p(n)s(n-J)]+
=o
∑ K=O∑i=Ik(^) • [^-k>(n-J) + p(n-k)p(n-J)+S(H-k)p(n-j)+p(n-k>("-J)i=o
(io:
With the hypothesis that the perturbation signal and the useful signal are not correlated, we assume that we can write
JV-I JV-I P ∑ [sin) pin - J) + pin)sin - J)] + ∑ ∑ [ay (k) - [s(n - k)pin - J) + pin - k)sin - j)]] « 0
B=O K=Oi=I
(11)
As a result equation Eq (10) is reduced to:
JV-I JV-I P
∑(sin)sin-J)+pin)pin-J))=-∑∑ay(k)-[sin-k)s(n-J)+pin-k)p(n-j)]
B=O K=Oi=I
(12: Assuming that the estimation s(n) from Eq (4) is close to the signal s(n) (s(n)~s(n)) we can write:
JV-I JV-I P
∑s(n)s(n-j) =-∑∑as(k)-s(n-k)s(n-j)
K=O K=Oi=I
(13:
=-∑as(k)-∑s(n-k)s(n-j) k=\ κ=0
In the same way, we can write:
JV-I JV-I P
∑P(n)p{n - j) = -∑ ∑ap(k) ■ p(n - k)p(n - j) κ=0 H=O k=\ P JV-I :i4)
= "∑ a p (k) -∑pin- k)p(n - j) k=\ κ=0
As a result Eq (11) leads to:
JV-I JV-I
-∑as{k)∑s{n-k)s{n- j)-∑ap(k)∑p(n-k)p(n- j) = k=\ κ=0 k=\ JV-I P (is:
∑∑ay(k) ■ [s{n - k)s(n - j) + p(n - k)p(n - J)] κ=0 k=\
Let's introduce the covariance functions of the useful signal s(n) and of the perturbation p(n)
N-I ru(J- j) = ∑u(n-i)-u(n- j), u = s,p,y (16;
B=O as well as the autocorrelation matrix of s(n) , p{n) and y{n)
Figure imgf000011_0001
'
For e.g. P=4 , T can be written as T =
Figure imgf000011_0002
Let's also introduce the LPC coefficients vector of the different signals considered ( s(n) p(n) andy(n) ) :
Figure imgf000011_0003
,P, u = s,p,y (18) Where τ stands for the transposition operator. With the new notation introduced Eq (14) becomes:
Figure imgf000012_0001
As a result the LPC coefficients of the useful signal can be obtained through:
Figure imgf000012_0002
Or equivalently :
As=Ay+Ts-\Tp[Ay-Ap] (2i:
The two last expressions show that the computation of the useful signal LPC is obtained when the following entities are known: the LPC of the perturbed signal (Ay), the LPC of the perturbation (Ap), the covariance matrix of the perturbation (Tp), and the inverse of the covariance matrix of the useful signal ( Ts ) . One can see the formula as a filter of the perturbed LPC Ay to obtain the useful LPC As , this filter depending on Ap , Tp and Ts .
Accordingly, the present invention proposes a method based on formula Eq. (20) -(21) or on any formula derived from this equation to obtain the LPC coefficients of the useful signal (A5), when the LPC coefficients of the perturbed signal (A ) are available. Eq. (20) /(21) require to know the LPC coefficients of the perturbation Ap , the correlation matrix of the perturbation Tp and the inverse of the correlation
matrix of the useful signal T~ . Generally, these entities are not directly available as we place our problem in a scheme where only the perturbed coefficients A are available. Accordingly Ap , Tp and Ts ~λ need to be estimated. It results that the invention can be seen as the generic process described below:
For each frame m, the LPC A are available.
- Estimation of the LPC of the perturbation Ap based on the LPC Ay . As an alternative, if more entities than Ay are available the estimation can be based on the additive information. For instance, when placing the method within a speech codec, other speech codec parameters can be used to obtain the estimation.
- Estimation of the correlation matrix of the perturbation Tp based on the LPC Ay . As an alternative, if more entities than Ay are available the estimation can be based on the additive information. For instance, when placing the method within a speech codec, other speech codec parameters can be used to obtain the estimation.
- Estimation of the correlation matrix of the useful signal Ts based on the LPC A . As an alternative, if more entities than Ay are available the estimation can be based on the additive information. For instance, when placing the method within a speech codec, other speech codec parameters can be used to obtain the estimation. Applying the filter defined in Eq (20) /(21) to get As .
This process can be applied on speech codec bitstream by applying the following steps: - For each frame m, extracting the LPC coefficients A from the speech codec bitstream.
- Applying the processed as described previously to get the useful signal LPCA5. - Exchanging the coefficients A with the useful one As
The solution has following advantages:
- The method makes it possible to obtain clean LPC coefficients when knowing perturbed ones. Any system based on linear prediction, computing coefficients A can use this solution to obtain the clean coefficients As
- A typical application of the method is processing noise reduction, echo reduction or reduction of any other perturbation on the LPC on speech. - When PCM samples are not available, but only codec parameters, the methods permits to re-construct the useful LPC parameters without the need to decode the bitstream to get the PCM data and apply classical noise reduction or echo cancellation. This is an alternative solution to existing prior art solutions.
- In the LPC analysis step (speech encoding) , such method can be applied in parallel to the LPC analysis. It can be seen as a kind of postfilter after the LPC analysis.
- The mathematical expression obtained in Equation (21) is relatively easy to implement. It requires of course estimations of certain entities like cross-correlation functions or LPC or the noisy signal, but such estimations are quite classical. The method is accordingly quite classical from a signal processing point of view, and possible to implement in real time applications.
The invention is demonstrated with help of the drawings. Figure 1 shows a signal transmission from a sender to a receiver in a telecommunication system.
Figure 2 is a flow chart for the modification of LPC coefficients according to a first embodiment.
Figure 3 shows a comparison of transfer function with non- modified LPC versus modified LPC coefficients.
Figure 4 shows a second embodiment for the modification of LPC coefficients.
Figure 1 shows an embodiment of a telecommunication system 1 in a signal transmission with modified LPC coefficients. The sender 2 generates the useful signal s (t) by talking.
Perturbations generate a perturbation signal p(t) with is added to the useful signal resulting in the signal y(t) . The signal y(t) is digitalized in the Analog-Digital-Converter (AD-Converter) 3 which generates a digital signal y(n) . The digital signal y(n) is encoded in the encoder to the signal ye (n) .
The encoding is done with the help of an LPC analysis. The encoded signal ye (n) is transmitted via the transmission block 5 to the decoder 6. The decoder 6 receives the signal ye ' (n) from the transmission block and decodes ye ' (n) to a digital signal yd(n). Depending on how the transmission block 5 is implemented, ye ' (n) is either equal or unequal to ye (n) . The transmission block 5 is e.g. a telephone switch, a router or a simple wire.
yd(n) is finally DA-converted by the DA-converter to ya(t) which is received as an analog signal by the receiver 8. In the embodiment of Figure 2, the modification of the LPC parameters is done in the transmission block 5, whereas in the embodiment of Figure 4, the encoder 4 directly modifies the LPC parameters .
Figure 2 is a flow chart for the modification of LPC coefficients within the transmission block 5. ye (n) is a bitstream including LPC coefficients. If the encoder uses the AMR codec, the LPC coefficients are transmitted as Line
Spectral Pair (LSP) . The frames of ye (n) also comprise the parameters pitch delay, fixed codebook index, fixed gain and adaptive gain. The bitstream is computed by the analysis of successive frames, each, each comprising a defined number of samples (generelly 160) . If the signal ye(t) is sampled at a frequency of 8 MHz, in the so-called narrow band, the number of LPC coefficients is chosen to 8 or 10 in current standardized codecs (AMR, EFR, FR) . In other words, the codec uses 8th respectively a 10th order linear prediction filter. In Eq. (1) k runs from 1 to 9 respectively from 1 to 11.
In the case of the so-called wide band, the sampling frequency is 16 kHz and the number of coefficients is preferably chosen to 16 in current standardized codecs (AMR- WB) .
The Figure 2 shows the flow chart where the LPC coefficients are extracted from the bitstream ye (n) . The bitstream is divided in the LPC coefficients and the rest of the bitstream, including the information needed to decode the residual waveform. Then the estimations of Ap , Tp and Ts ~ are applied, taking into account the LPC coefficient as well as eventually additive information from the bitstream. In an embodiment, Ap is generated by the help of a Voice
Activity Detection (VAD) . Voice Activity Detection is known in the art. Here, the output of the VAD generates zero if no voice signal is detected in the bitsteam, else the VAD outputs a one. Ap(m) is calculated by the following algorithm, whereby m is the index of the frame or subframe.
if VAD = 0, Ap(m) = aAy(m) + (\-a)Ap(m-\) if VAD = 1, Ap(m) = Ap(m-\) a is a fixed parameter being heuristically chosen, whereby 0<α<l .
To estimate Tp , the perturbation is assumed to be white noise. Accordingly, the autocorrelation matrix Tp has the following form:
Figure imgf000017_0001
Depending on the output of VAD, Ep (m) is calculated by the equations if VAD = 0, Ep(m) = βEy(m) + (l-β)Ep(m-l) if VAD = 1, Ep(m) = Ep(m-\),
wherein by E (m) is the energy of the signal y(n), m indicates a frame or subframe and β is a fixed parameter being heuristically chosen, 0</?<l.
In this embodiment, Ts is estimated by the equation: r =r -r .
Ty may be calculated with the help of Eq. 15 if the data stream is decoded.
In this embodiment, the bitstream of the encoded signal has to be decoded to make the estimation of Tp , Ts .
Alternatively, the estimation of this matrices and vectors can also be done on the basis of the codec parameters of the signal ye (n) by interpreting the fixed gain and the adaptive gain.
After the estimation of As , Tp , Ts , the clean LPC coefficients As are generated by one of the equations 21 or 22. It should be noticed that the calculated clean LCP As are an estimation of a LPC of the useful signal s (n) . Accordingly, the calculated LPC As are as good as the estimations for
Figure imgf000018_0001
The filter on the LPC A coefficients is applied to get the clean LPC As and finally the LPC are replaced in the bitstream by changing each frame by the use of the clean LPC parameters As . The frames are modified sequentially and sent to the decoder as signals ye ' (n) .
This method of improving the speech signal quality can be done anywhere in the path between the encoder and decoder. For example, in telecommunication systems, the method can be applied in the terminal of the sender, in the terminal of the receiver or in one of the routers telephone switches or gateways between different networks. The use of the modified LPC coefficients improves the quality of the received signal which is demonstrated with the help of Figure 3.
Figure 3 shows a comparison of transfer functions with non- modified LPC versus modified LPC coefficients. The synthesis LPC filter function can be described by the filter transfer function H(f) in the frequency domain. The graph of Fig. 3 shows a functions H(f) dependent on the frequency f for a non-noisy LPC function and, with the dashed line, for a noisy LPC filter. The transfer function of the noisy LPC filter, in our case a non-modified LPC, has more energy but is smoother. Using a LPC that was generated on the basis of a noisy signal worsens the quality of the speech. Hence, the modification of the LPC to a clean LPC make it easier for the receiver to understand the received speech and enhance the clarity of the speech.
If only the LPC 's are modified, the received speech still includes noise. Therefore, as an option, the flow chart of
Figure 2 may be extended by an additional step which reduces noise on the rest bitstream. This noise reduction is performed after the estimation of Ap , Tp and T ~ and before the generation the new frames of the bitstream of the signal.
One examplary noise reduction technique for the rest bitstream is the method for reducing noise on the codec parameters pitch gain and codebook gain described in the above-mentioned "Compressed Domain Noise Reduction and Echo Suppression for Network Speech". Figure 4 shows a second embodiment of the modification of LPC coeffients. In this case, the function is included in the encoder 4.
In speech encoding, like typically AMR, AMR-WB, G723, G729, the LPC coefficient are computed by an analysis of successive weighted frames. Typically, the Levinson-Durbin algorithm permits to get the LPC coefficient from the sample y(n) of the analysis frame. Our method maybe placed as a postfilter of the computation blocks of the LPC analysis. In this scenario, it enhances the LPC coefficient by reducing the influence of the noise. The needed estimations (A^, Tp and
Ts ) may be done by using the LPC coefficients but also some additive information of sample y(n) as depicted in Fig 4. Finally, the filter of Eq (20) /(21) is applied on the perturbed coefficients to get the enhanced ones.
In the embodiment of Figure 4, the method of improving the speech quality is performed within the encoder. The encoder receives the samples from the A/D-converter . The samples are organized as frames. After windowing a frame, the LPC analysis outputs the LCP coefficients Ay . The parameters Ap ,
Tp and Ts ~ are estimated like one of the embodiments described above. LPC coefficients Asare calculated by one of the equations (21) and (22) . The encoding of the frame is done with A . Reference number list
1 telecommunication system 2 sender
3 A/D-converter
4 encoder
5 transmission block
6 decoder 7 D/A-converter
8 receiver

Claims

Patentansprϋche / Patent claims
1. Method for transmitting a digital signal y(n), y(n) comprising a useful signal s (n) and a perturbation signal p (n) , the method comprising the steps of:
- receiving the Linear Prediction Coefficients (LPC) A of a bitstream ye (n) , ye (n) being an LPC-encoded signal of y (n) ; - estimating the autocorrelation matrix Ts of the useful signal s (n) , of the autocorrelation matrix Tp of the perturbation signal p (n) and the LPC Ap of the perturbation signal p (n) ;
- calculating a modified LPC As by using A and the estimated Fs, Yp, Ap ;
- outputting a modified data stream ye ' (n) including the modified LPC As .
2. Method according to claim 1, wherein As is calculated by A5=F5 .[(F5 +Tp)- Ay — Tp -Ap\ .
3. Method according to claim 1, further comprising the step:
- receiving the residual signal of the encoded signal ye (n) and wherein in the step of
- estimating Fs, Tpl Ay the estimated values for at least one of the parameters Ys, Tp, Ay are calculated from the residual signal of ye (n) .
4. Method according to one of the preceding claims, further comprising the steps:
- performing a noise reduction on the residual signal of the encoded signal ye (n) .
5. Method according to one or the claims 1 to 4, in which instead of
- receiving the Linear Prediction Coefficients (LPC) A of the signal ye (n) , the method comprises the steps of
- receiving the digital signal y(n) and
- calculating Linear Prediction Coefficients (LPC) Ay by encoding the signal y(n) .
6. Method according to one of the preceding claims characterized in that the method is performed in a telecommunication system.
7. Digital signal transmission apparatus which performs a method according to one of the claims one to 6.
8. Digital signal transmission apparatus according to claim 7, comprising a Digital Signal Processor (DSP).
9. Digital signal transmission apparatus according to claim 7 being a telecommunication terminal.
PCT/EP2007/063598 2007-01-15 2007-12-10 Disturbance reduction in digital signal processing WO2008086920A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
EP07000716A EP1944761A1 (en) 2007-01-15 2007-01-15 Disturbance reduction in digital signal processing
EP07000716.6 2007-01-15

Publications (1)

Publication Number Publication Date
WO2008086920A1 true WO2008086920A1 (en) 2008-07-24

Family

ID=38007980

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2007/063598 WO2008086920A1 (en) 2007-01-15 2007-12-10 Disturbance reduction in digital signal processing

Country Status (2)

Country Link
EP (1) EP1944761A1 (en)
WO (1) WO2008086920A1 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9224403B2 (en) 2010-07-02 2015-12-29 Dolby International Ab Selective bass post filter

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
MX2018003529A (en) * 2015-09-25 2018-08-01 Fraunhofer Ges Forschung Encoder and method for encoding an audio signal with reduced background noise using linear predictive coding.

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2002054744A1 (en) * 2000-12-29 2002-07-11 Nokia Corporation Audio signal quality enhancement in a digital network
WO2002080149A1 (en) * 2001-03-30 2002-10-10 Telefonaktiebolaget Lm Ericsson Noise suppression

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2002054744A1 (en) * 2000-12-29 2002-07-11 Nokia Corporation Audio signal quality enhancement in a digital network
WO2002080149A1 (en) * 2001-03-30 2002-10-10 Telefonaktiebolaget Lm Ericsson Noise suppression

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
CHANDRAN R ET AL: "Compressed domain noise reduction and echo suppression for network speech enhancement", CIRCUITS AND SYSTEMS, 2000. PROCEEDINGS OF THE 43RD IEEE MIDWEST SYMPOSIUM ON AUGUST 8-11, 2000, PISCATAWAY, NJ, USA,IEEE, vol. 1, 8 August 2000 (2000-08-08), pages 10 - 13, XP010558066, ISBN: 0-7803-6475-9 *

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9224403B2 (en) 2010-07-02 2015-12-29 Dolby International Ab Selective bass post filter
US9343077B2 (en) 2010-07-02 2016-05-17 Dolby International Ab Pitch filter for audio signals
US9396736B2 (en) 2010-07-02 2016-07-19 Dolby International Ab Audio encoder and decoder with multiple coding modes
US9552824B2 (en) 2010-07-02 2017-01-24 Dolby International Ab Post filter
US9558753B2 (en) 2010-07-02 2017-01-31 Dolby International Ab Pitch filter for audio signals
US9558754B2 (en) 2010-07-02 2017-01-31 Dolby International Ab Audio encoder and decoder with pitch prediction
US9595270B2 (en) 2010-07-02 2017-03-14 Dolby International Ab Selective post filter
US9830923B2 (en) 2010-07-02 2017-11-28 Dolby International Ab Selective bass post filter
US9858940B2 (en) 2010-07-02 2018-01-02 Dolby International Ab Pitch filter for audio signals
US10236010B2 (en) 2010-07-02 2019-03-19 Dolby International Ab Pitch filter for audio signals
US10811024B2 (en) 2010-07-02 2020-10-20 Dolby International Ab Post filter for audio signals
US11183200B2 (en) 2010-07-02 2021-11-23 Dolby International Ab Post filter for audio signals
US11610595B2 (en) 2010-07-02 2023-03-21 Dolby International Ab Post filter for audio signals
US11996111B2 (en) 2010-07-02 2024-05-28 Dolby International Ab Post filter for audio signals

Also Published As

Publication number Publication date
EP1944761A1 (en) 2008-07-16

Similar Documents

Publication Publication Date Title
US7539615B2 (en) Audio signal quality enhancement in a digital network
US20210035596A1 (en) Speech signal cascade processing method, terminal, and computer-readable storage medium
US20120263317A1 (en) Systems, methods, apparatus, and computer readable media for equalization
US7558729B1 (en) Music detection for enhancing echo cancellation and speech coding
US6694018B1 (en) Echo canceling apparatus and method, and voice reproducing apparatus
US20060215683A1 (en) Method and apparatus for voice quality enhancement
US20070160154A1 (en) Method and apparatus for injecting comfort noise in a communications signal
US20110054889A1 (en) Enhancing Receiver Intelligibility in Voice Communication Devices
US20060217969A1 (en) Method and apparatus for echo suppression
US20060217972A1 (en) Method and apparatus for modifying an encoded signal
EP3353783A1 (en) Encoder and method for encoding an audio signal with reduced background noise using linear predictive coding
JP2010503325A (en) Packet-based echo cancellation and suppression
US8874437B2 (en) Method and apparatus for modifying an encoded signal for voice quality enhancement
US20060217983A1 (en) Method and apparatus for injecting comfort noise in a communications system
US20060217988A1 (en) Method and apparatus for adaptive level control
US20060217970A1 (en) Method and apparatus for noise reduction
US6718036B1 (en) Linear predictive coding based acoustic echo cancellation
US20030065507A1 (en) Network unit and a method for modifying a digital signal in the coded domain
CA2244008A1 (en) Nonlinear filter for noise suppression in linear prediction speech pr0cessing devices
US20060217971A1 (en) Method and apparatus for modifying an encoded signal
EP1944761A1 (en) Disturbance reduction in digital signal processing
JP2018511086A (en) Audio encoder and method for encoding an audio signal
EP1521241A1 (en) Transmission of speech coding parameters with echo cancellation
EP1521240A1 (en) Speech coding method applying echo cancellation by modifying the codebook gain
Beaugeant et al. Gain loss control based on speech codec parameters

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: 07848027

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: 07848027

Country of ref document: EP

Kind code of ref document: A1