Telephonic communications – Echo cancellation or suppression – Residual echo cancellation
Reexamination Certificate
2000-07-28
2003-07-22
Isen, Forester W. (Department: 2644)
Telephonic communications
Echo cancellation or suppression
Residual echo cancellation
C379S406010, C379S406020, C379S406030, C379S406060, C379S406080, C379S406090
Reexamination Certificate
active
06597787
ABSTRACT:
This application claims priority under 35 U.S.C. §§119 and/or 365 to 199 35 808.7 filed in Germany on Jul. 29, 1999; the entire content of which is hereby incorporated by reference.
FIELD OF THE INVENTION
The invention relates to an echo cancellation device for canceling echoes caused by a coupling of a reception signal received by a receiving unit of a transceiver unit of a telecommunication system to a transmitting unit of the transceiver unit. In particular, the invention eliminates echoes which are introduced in the transmitting path of the transceiver unit as a result of an acoustic coupling between a loudspeaker of the receiving unit to a microphone of the transmitting unit.
More specifically, the echo cancellation device is intended to eliminate so-called residual echoes in the transmitting path which remain in the output of a conventional echo canceller after a main echo cancellation has been carried out.
BACKGROUND OF THE INVENTION
FIG. 1
shows in connection with
FIG. 2A
a block diagram of a conventional echo canceller EC of a transceiver unit TRU of a telecommunication system TELE. Via an antenna ANT and an antenna switch SW a signal RFE′″ is input and processed by a receiving unit RX. A receiver circuitry RCRT and a decoder DECOD contain all the high frequency and low frequency circuits for providing a reception signal RFE to a loudspeaker SP via a D/A-converter and to the echo canceller EC. In the low frequency path of the receiving unit RX the speech decoder DECOD recomposes speech from the information contained in the signal RFE′″ (see FIG.
1
). This recomposing of speech will be explained with more detail with reference to
FIG. 4
which shows a schematic block diagram of the speech decoder DECOD. Hereinafter, the signal RFE received from a far end transceiver unit will also be called the “far end signal” whilst the signal TFE provided by the near end transceiver unit to the far end transceiver unit will be denoted as the “transmitted near end signal”.
As in particular shown schematically in
FIG. 2A
, the far end signal RFE is emitted from the loudspeaker SP of the transceiver unit TRU and is acoustically coupled to the transmitting unit TR, in particular to the microphone MC thereof. Also other coupling effects are conceivable, i.e. through a parasitic electrical coupling between the receiving and the transmitting units RX, TR. Thus, the far end signal emitted from the loudspeaker SP together with the microphone MC form a closed loop system causing the far end signal RFE to be transmitted back to the far end transceiver unit.
In most telecommunication systems TELE, in particular in a global system for mobile communications (Global System for Mobile Communication GSM), the transmitted signal TNE′, TFE will be delayed, such that the user of a far end transceiver unit will perceive this as an echo. In this connection it should be noted that the teachings disclosed herein are not particularly limited to a mobile radio communication system but also apply to other communication systems where two transceiver units transmit and receive speech. Therefore, the radio transmission via an antenna ANT is only one example of such telecommunication systems.
Due to the acoustic and/or electrical coupling effect, a portion of the far end signal will always be present in the transmitting path independently as to whether or not the user of the near end transceiver unit actually speaks into the microphone MC or not. This aspect as to whether speech is present or not will be investigated with more details below.
Prior Art I: Residual Echo Cancellation
To eliminate the far end signal being transmitted to the far end transceiver unit, an echo cancellation device EC comprising a transfer function estimator EST, H and a subtractor ADD is used, cf. FIG.
2
B. Basically, the transfer function estimator EST, H is adapted to estimate the coupling transfer function H from the receiving unit RC to the transmitting unit TR and for processing the reception signal RFE with said estimated coupling transfer function H. In particular, if the acoustic coupling is considered, the transfer function estimator EST, H estimates the acoustic transfer function from loudspeaker SP to the microphone MC. The filter output signal. RFE′ is subtracted by the subtractor ADD from the transmission signal TNE which includes an echo signal due to the acoustic and/or electric coupling of the received signal RFE to the transmitting unit. Ideally, the use of the transfer function estimator and the subtractor should be enough to completely eliminate the occurrence of the reception signal RFE in the output signal TNE′ from the echo canceller EC.
However, in practice the main or basic echo cancellation by using the transfer function estimator and the subtractor cannot remove the returning signal completely. The reason for this is that the transfer function estimator H, EST cannot perfectly estimate the transfer function, in particular the transfer function of the acoustic coupling between the loudspeaker SP and the microphone MC. Consequently, some parts of the received far end signal RFE will still be present in the signal TNE′ transmitted to the far end transceiver unit. In the far end transceiver unit such remaining parts will still be perceived as an echo. Since a main echo cancellation has already removed some of the main echoes, the remaining parts of the far end signal are called “residual echoes”. Therefore, additional signal processing has to be applied to the residual signal TNE′ and in the context of conventional echo cancellation this additional processing is called “residual echo cancellation”. Thus, in some conventional echo cancellation devices an additional residual echo suppression device is used for suppressing residual echoes in the subtractor output signal TNE′. This will be considered below with reference to some examples of the published prior art.
Prior Art II: GSM Speech Coding/Decoding
In modern mobile communication systems, i.e. GSM, the voice signal TNE′ of
FIG. 1
, is not transmitted as a representation of the voice signal amplitudes. Instead the voice signal is coded and in GSM the speech coding is based on a model for speech generation. Commonly used methods to model speech are described in L. R. Rabiner and R. W. Schafer,
Digital Processing of Speech Signals,
Prentice Hall, Englewood Cliffs, N.J., 1978. In particular, a model which models the excitation signal and the vocal tract of the speaker is often used in signal processing. This model is defined by two types of excitation signals and a filter. The two excitation signals correspond to:
1) a pulse train used for voiced speech, e.g. the sound “a”;
2) a white noise used for unvoiced speech, e.g. the sound “s”.
The used filter models the vocal tract and it is convenient to use an AutoRegressive (AR) filter. By using the speech model it is possible to create an artificial voice. Actually, the voice will sound unnatural due to the excitation signals. However, if the excitation is chosen with care, more natural sounding speech can be produced.
Typically, speech modelling is used in speech coders, e.g. in the Full Rate (FR) coder in GSM. The FR coder is known as a Regular Pulse Excitation-Long Term Prediction (RPE-LTP) coder and is described in for example the GSM specification GSM 06.10. A simplified description, see
FIG. 3
, of the FR coder is as follows:
A frame of input samples TNE′, in GSM one frame consists of 160 samples, is presented to the coder input, e.g. in the form of the signal TNE′ output by the echo canceller EC. The input is used so as to determine an AR model, in
FIG. 3
represented by COD-AR. This is accomplished by exploiting the Toeplitz structure of the TNE′ correlation matrix, i.e. using a Schür recursion as described in J. G. Proakis and D. G. Manolakis:
Digital signal processing: principles, algorithms and applications,
Macmillan, publishing company, New York, 2nd edition, 1992. This recursion results in a set of coefficients
Lindgren Ulf
Misra Mohan
Philipsson John
Burns Doane Swecker & Mathis L.L.P.
Isen Forester W.
Singh Ramnandan
Telefonaktiebolaget L M Ericsson (publ)
LandOfFree
Echo cancellation device for cancelling echos in a... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Echo cancellation device for cancelling echos in a..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Echo cancellation device for cancelling echos in a... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3100845