Telephonic communications – Echo cancellation or suppression – Using digital signal processing
Reexamination Certificate
1998-04-10
2003-03-25
Tsang, Fan (Department: 2644)
Telephonic communications
Echo cancellation or suppression
Using digital signal processing
C379S392000, C379S390010
Reexamination Certificate
active
06539091
ABSTRACT:
BACKGROUND OF THE INVENTION
Field of the Invention
The invention relates to an apparatus for sidetone (noise) damping which can be used in telephones configured for hands-free operation. Telephones intended for connection to the public telephone system must meet certain specifications. One of these specifications involves sidetone damping, that is, the damping of the echo from a remote conversation partner. In International Telecommunication Standardization Sector (ITU-T) Recommendation G. 164, Acoustic Echo Controller, Helsinki, March 1993, Section 5.4.1., echo damping of 40 to 45 dB is called for in the case of one individual talking (that is, no more than one of the two partners in the conversation is speaking). In a conversation using a handset these damping values are easily met because of the extensive acoustical decoupling between the receiver and the microphone capsule. In a hands-free telephone setup that has no handset such acoustical decoupling is not possible. Special measures must then be taken to achieve the requisite sidetone damping.
In most commercially available telephones configured for hands-free operation, there is a damping device (level scales, adaptive loss control, level cradle), which depending on the conversation status performs the requisite damping in either the transmission branch or the reception branch. Such a telephone can be found in the ARCOFI component PSB 2163 made by Siemens. Difficulties arise in hands-free operation based on level scales especially if the conversation partners try to interrupt one another or talk at the same time. Then the introduced damping has a negative effect and can result in isolated speech portions being “chopped off” or entirely suppressed. For many users, this unsatisfactory performance of the level scales during crosstalk has prevented commercial acceptance of the hands-free telephone configuration.
A completely different principle for achieving the requisite echo damping is provided by digital echo compensators. Digital echo compensators generate estimated values of the sampled values of the acoustical echo in the microphone signal. By subtracting these estimated values from the sampled values of the actual microphone signal, the proportion of the echo in the microphone signal is reduced. The use of digital filters for echo compensation in hands-free telephone systems is described in T. Becker, E. Hänsler and U. Schultheis, Probleme bei der Kompensation Akustischer Echos [Problems in Acoustical Echo Compensation], Frequenz 6 (36) pp. 142-148 (1984). Echo compensators are configured as adaptive filters, with which one seeks to simulate the transmission environment of the room in which the telephone set is located as accurately as possible and to follow along with changes in the transmission environment that are caused by user movements or by air and temperature fluctuations. The adaptation of the filter coefficients is allowed to occur only if a speech signal on the part of the remote subscriber is present. As soon as the local partner also speaks, the echo compensator can diverge. In crosstalk (doubletalk), the adaptation must therefore be stopped or at least sharply slowed down. Due to a lack of system calibration, the echo damping of an echo compensator can be slight at certain times. These and other peripheral conditions (high computer capacity and memory required, space required, and heat development in the circuit) mean that under realistic conditions echo damping of approximately 20 to 25 dB cannot be undershot.
In order to be able to assure the requisite damping, as a rule, an echo compensator is operated only in conjunction with a level scales. The damping stroke of the level scales can then be reduced by the echo damping already performed by the echo compensator, so that the switching behavior of the level scales is no longer perceived as being so irritating.
From the prior art article by P. Heitkämper, Freisprechen mit Verstärkungssteuerung und Echokompensation [Hands-Free Telephone with Gain Control and Echo Compensation], Fortschritt-Berichte, Reihe 10, Nr. 380 [Progress Reports, Series 10, No. 380], VDI [Association of German Engineers], Germany, 1995, a hands-free system and process is described that is based on the combination of gain control and an echo compensator. The system gain control uses a nonlinear characteristic curve that serves to amplify all the input values on the transmission side, whose average short term magnitude is above a certain threshold value, in such a way that their average power assumes an intended desired level (compensation range). Input values whose average power is less than the threshold value are considered to be background noise or echo and are damped as a function of their average power. The average short term magnitude of the input data is ascertained by smoothing the magnitudes of the input data with a nonlinear filter, whose time constants are selected such that they are capable of following rising magnitudes in the input signal faster than falling magnitudes. The threshold value, which determines whether the input data will be treated like a desired speech signal or like an undesired echo or background noise, is adapted adaptively to the background noise prevailing in the room, to the echo level, and to the level of the local conversation partner. For adapting the threshold value, long-term average values for the local and remote background noise and speech levels, a correlation measurement for speech activity detection, and an estimated value for the speaker-microphone coupling factor (power ratio between the received signal and the transmitted signal) are determined.
The echo compensation uses an NLMS algorithm (normalized least mean square algorithm), which is described in B. Widrof and M. E. Hoff, Adaptor switching circuits, IRE Westcon Convention CONV. Rec., Part 4, pp. 96-104, 1960. For the NLMS algorithm, a simple increment width controller sets the adaptation increment width to zero if excitation from the remote conversation partner is inadequate or if crosstalk is detected. In all other cases, the adaptation increment width is set to a constant value between 0 and 1. The filtering is done merely with subsampling of the transmitted frequency band, for the sake of reducing the effort. Adequate excitation is ascertained by using a comparison between the average short term magnitude of the excitation and the long term average magnitude of the remote background noise level. Crosstalk is detected if the average short term magnitude of the transmitted signal rises above the value of the expected echo (which is equivalent to the average short term magnitude of the received signal, multiplied by the speaker-microphone coupling factor, multiplied by the safety factor). The speaker-microphone coupling factor in turn is only estimated if a previously calculated correlation measurement has undershot a certain threshold.
Using the nonlinear characteristic curve in the gain control cannot, in the method described, assure that the aforementioned ITU-T Recommendation G. 164 can be adhered to in all conversation situations using the hands-free method.
Particular methods of the type described in the above publication by P. Heitkämper in VDI Fortschritt-Berichte have also been disclosed in Non-Prosecuted, German Patent Applications DE 42 29 910 A1 and DE 43 05 256 A1.
SUMMARY OF THE INVENTION
It is accordingly an object of the invention to provide an apparatus for sidetone damping, which overcomes the herein-mentioned disadvantages of the heretofore-known devices and methods of this general type, and in which the ITU-T Recommendation is adhered to such that the “chopping off” of individual speech passages is avoided.
With the foregoing and other objects in view there is provided, in accordance with the invention, an apparatus for acoustical sidetone damping, including: a reception branch having a speaker and receiving an input signal; a transmission branch having a microphone for transmitting an output signal; an adaptive echo c
Harold Jefferey
Infineon - Technologies AG
Tsang Fan
LandOfFree
Apparatus for sidetone damping does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus for sidetone damping, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus for sidetone damping will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3081963