Wave transmission lines and networks – Plural channel systems – Having branched circuits
Reexamination Certificate
2000-07-13
2004-09-14
Summons, Barbara (Department: 2817)
Wave transmission lines and networks
Plural channel systems
Having branched circuits
C333S133000, C333S020000, C333S02800T, C333S166000, C333S175000, C333S186000, C704S271000, C381S023100, C381S314000, C381S316000
Reexamination Certificate
active
06791433
ABSTRACT:
TECHNICAL FIELD
The invention concerns signal processing systems comprising resonators arrays and schemes for signal processing.
BACKGROUND OF THE INVENTION
The market of computing expands rapidly in pervasive computing devices (also referred to as tier-
0
devices). As these devices will be part of almost every aspect of our life, and as the functions of these devices will become increasingly complex, it is mandatory that the interaction between man and machine mimics the ways humans interact with each other. Thus there is a need of simple signal processing techniques, e.g., for the processing of voice or other audio signals.
There are many other areas where there is a demand for better signal processing systems. Examples are hearing aids, speaker recognition systems, noise suppression systems, and systems that allow a speaker to control certain functions of a computer, device, vehicle, machine, or apparatus, and so forth.
Signal processing typically involves filtering of an input signal in the frequency domain. In the standard approach, the input signal is first transformed in to the frequency domain using well known Fast Fourier Transform (FFT) algorithms. In the frequency domain, implementation of the filter operation is accomplished by a multiplication of the filter response function with the transformed input signal. A filter ed time domain output signal is then recovered by means of a subsequent inverse FFT operation. Although extremely efficient algorithms are known from implementation this approach it suffers from a number of fundamental deficiencies. The FFT transformation is a non-local operation in the time domain. This means that an intrinsic time delay is introduced which is inversely proportional to the frequency resolution of the transformation. In order to avoid disturbing echoing effects the time delay should typically not exceed 10
−2
s in audio applications which corresponds to a frequency resolution of 100 Hz, at best. In order to achieve sufficient frequency resolution for a satisfactory synthesis of the filtered signal one must resort to complex phase analysis methods which substantially increase the computational effort.
Standard signal processing techniques are computationally expensive and are thus not suited for use in voice recognition or speaker recognition as implemented in some tear-
0
devices such as smart cards.
Similar problems occur in today's hearings aids, where specified spectral regions are amplified. For these applications the time delay is particular detrimental, because here the original and the processed signal overlap, and thus smear out the subtle time information.
It is a disadvantage of known systems that the time resolution is not high enough to provide reliable cues for use by speech recognition systems.
It is an object of the present invention to provide an improved signal processing approach.
It is an object of the present invention to provide an improved signal processing approach which is suited for use in hearing aids.
It is an object of present invention to provide an improved signal processing approach for use in speech recognition or speaker recognition systems.
It is an object of the present invention to provide an improved signal processing approach for processing acoustic signals.
SUMMARY OF THE INVENTION
The present invention concerns a method, signal processing system, a computer program element, and a computer program product as claimed in the claims.
Advantages of the present invention are either explicitly addressed in the detailed description or are obvious from the disclosed.
REFERENCES:
patent: 4823795 (1989-04-01), van den Honert
patent: 5736909 (1998-04-01), Hauser et al.
patent: 5748838 (1998-05-01), Stevens
patent: 5856722 (1999-01-01), Haronian et al.
patent: 6243671 (2001-06-01), Lago et al.
patent: WO 97/13127 (1997-04-01), None
Bloechl Peter
Duerig Urs
Folini Oliver
Dougherty Anne V.
International Business Machines - Corporation
Jennings Derek S.
Summons Barbara
LandOfFree
Signal processing by means of resonators does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Signal processing by means of resonators, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Signal processing by means of resonators will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3254748