Electrical audio signal processing systems and devices – Dereverberators
Reexamination Certificate
2002-10-29
2004-04-06
Mei, Xu (Department: 2644)
Electrical audio signal processing systems and devices
Dereverberators
C379S406020, C379S406030
Reexamination Certificate
active
06718041
ABSTRACT:
FIELD OF THE INVENTION
The invention relates to the field of acoustics in air and it applies in particular to systems for wiring halls and rooms for sound.
The present invention relates to an echo reduction method implemented with a multi-sensor sound pickup device constituting an antenna and a sound playback device. The invention also relates to echo reduction apparatus designed to operate with a sound playback device to implement the method of the invention. The frequency domain to which the invention applies is the speech domain and more particularly that of low frequency speech signals.
Description of the Prior Art
A most advantageous application of the invention lies in reducing the echo in a sound pickup device in a teleconferencing system, thereby making so-called “hands-free” communication possible without needing to use an offset microphone.
Acoustic echo is a major obstacle to proper operation of hand-free communications terminals. This acoustic echo is the result of the sensors of the sound pickup device capturing a portion of the signal emitted by the transducers of the sound playback device. The sensors and the transducers used for acoustics in air are respectively microphones and loudspeakers.
In a hands-free communications terminal, microphones are acoustically coupled, and possibly also mechanically coupled, with the sound playback system, either directly when the microphones and the playback system are in the same housing, or else indirectly as is the case for a video conference terminal placed on a TV monitor. The resulting echo comprises two portions:
the room response; and
direct coupling between the loudspeaker and the microphone.
The response of the room can generally be processed effectively by conventional echo controlling techniques such as echo cancelling or gain variation, and in particular because of the relatively low level of this portion of the echo. The same does not apply to the direct coupling which results both from the sound path through the air and also from vibration propagating through the shell of the communications terminal, and possibly also from any resonances of cavities or mechanical elements when set top boxes are used. This portion of the echo is generally at a level which is 6 dB to 20 dB higher than the level of the local speech picked up by the microphones, and it gives rise to a “howl-around” effect unless processing is applied.
If known echo cancelling techniques do not enable this unwanted effect to be processed in satisfactory manner, then a large variation in gain must be applied in order to prevent howl-around starting. This variation in gain must apply firstly during the stage in which the echo canceller is converging and secondly following any variation in the echo path or during moments of double talking. Because of this constraint, the communications terminal is not very interactive and the user can have the feeling that the hands-free function does not work well.
To mitigate that drawback, there exists several techniques for reducing direct coupling.
A first technique sets out to reduce mechanical coupling. The solution normally applied consists in using damping materials such as foams or rubber to isolate the microphone from the shell. Those materials have the effects of eliminating or greatly reducing loudspeaker-generated vibration in the shell, and also propagation thereof to the microphone.
The mechanical decoupling solution is effective in reducing coupling by vibration, but industrially speaking it is expensive. Furthermore, it does not reduce the acoustic coupling that can be large if the microphone is placed close to the loudspeaker, as is the present trend because of the desire for communications terminals to be small in size. Furthermore, when the playback system forms an integral portion of an assembly that also includes the sound pickup device, as is the case of a television monitor including loudspeakers and microphones or of a monitor having placed thereon a device that includes microphones, it is not possible to envisage isolating the loudspeakers from the shell of the assembly.
A second technique consists in using a digital compensation filter whose response is obtained by calculating the inverse of the impulse response of the mechanical and acoustic coupling between the microphone and the loudspeaker.
In theory, the signal from the loudspeaker due to the echo is cancelled from the output of the compensation filter. In practice, that technique does not give satisfaction the characteristics of the coupling change, even if only to a very small extent, for example due to the communications terminal being assembled and/or disassembled or due to a change in the characteristics of the microphones. Furthermore, that technique is unsuitable if the disturbances are of a non linear nature, i.e. if the disturbances cannot be modelled by a convolution product between the signal coming from the loudspeaker and a filter impulse response. Furthermore, the behavior of transducers is rarely linear, since they are generally subject to distortion and/or to saturation, both of which are typical examples of non-linear operation. Finally, in mass production, coupling is necessarily different from one communications terminal to another, for example because the loudspeakers and microphones that are used differ slightly from one another. Finally, those limitations make the filter compensation technique relatively ineffective.
Other systems exist for weighting microphones based on uniformly distributing microphones around a loudspeaker and amplitude-weighting the microphones and/or phase-shifting the microphones. Such systems constitute the subject matter of French patent No. 93/020504. Those systems are for use in group communications terminals possessing unusual geometrical symmetry due to the way they pick up sound omnidirectionally. The effectiveness of those systems is highly sensitive to mismatches between the microphones or the acquisition channels. Consequently, those systems require components to be sorted and microphones to be calibrated accurately, and such matching can become lost over time. In addition, adding a display screen, a keypad, or a swivel-mounted camera can break the symmetry between the loudspeakers and the microphones. Furthermore, when the number N of microphones exceeds two, such devices are not optimal since N degrees of freedom are available for satisfying two constraints, namely fixing gain in a given direction and cancelling direct coupling. Finally, since the signal as picked up corresponds to all of the contributions from each of the microphones and not only the contribution from the wanted source, and since these contributions are mutually phase-shifted and possibly of different amplitudes, the spectrum of the resulting signal is degraded at high frequencies.
Another system intended more for individual applications, as disclosed in French patent No. 98/14321 for example, offers the advantage of taking account of the specific features of the final terminal, i.e. of the communications terminal after the system has been integrated therein. That system makes use of two microphones positioned at different distances from the loudspeaker. The contributions from the two microphones are combined by weighting so as to cancel the direct coupling wave. The different weightings of the two signals coming from the two microphones make it possible to pick up useful sources situated in a far field. The path filter applied to the second microphone can be calculated by inverting the response of a measurement of the coupling between the microphone and the loudspeaker. Such calculation is generally sufficient in most situations since the microphones are in the direct field of the loudspeaker. Nevertheless, that solution requires there to be an amplitude difference of sufficient magnitude in the coupling waves between the two microphones. To obtain such a difference, it is necessary to position the microphones sufficiently close to the loudspeaker which gives rise to conditions that are very unfavorable in terms
Cohen & Pontani, Lieberman & Pavane
France Telecom
Mei Xu
LandOfFree
Echo attenuating method and device 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 attenuating method and device, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Echo attenuating method and device will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3263703