Data processing: structural design – modeling – simulation – and em – Modeling by mathematical expression
Reexamination Certificate
1998-12-02
2003-04-01
Broda, Samuel (Department: 2123)
Data processing: structural design, modeling, simulation, and em
Modeling by mathematical expression
C703S005000, C704S207000, C704S270000, C704S258000, C704S262000, C702S104000
Reexamination Certificate
active
06542857
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates generally to systems and methods for characterizing, synthesizing, and/or canceling out acoustic signals from inanimate sound sources, and more particularly for using electromagnetic and acoustic sensors to perform such tasks.
2. Discussion of Background Art
Sound characterization, simulation, and cancellation are very important ongoing fields of research and commercial practice. Inanimate sound sources range from pleasant sounding musical instruments to very harsh and possibly harmful sounds from engines, air ducts, machines, and/or heavy equipment. Modern recording systems analyze and characterize inanimate sound sources, synthesizers use this characterized data to attempt to mimic various musical instruments, and noise cancellation technology uses prior analyzed data to reduce undesired sound levels.
Two prior art methods are in use today for analyzing and synthesizing sounds. The first method records long time duration segments of a sound, then divides them into shorter “segments.” During synthesis, the segments are arranged and concatenated as needed to synthesize a desired sound sequence. Such methods are often used to simulate musical instruments, as described in U.S. Pat. No. 5,111,727 entitled, “Digital sampling instrument for digital audio data” by D. P. Rossum. The second method of sound analysis and synthesis involves using measured or simulated sound system elements, such as sine waves, which are processed and modified for play back as desired. A prior art analog system of this nature is described in U.S. Pat. No. 4,018,121 entitled, “Method of synthesizing a musical sound” by J. M. Chowning. However, a shortcoming of these two prior art methods is that they do not use excitation source information to analyze, synthesize, and cancel sounds that inanimate objects make.
FIG. 1
is a dataflow diagram of a third prior art sound analysis and synthesis system. The third system is typical of those currently used in digital musical synthesizers and is further described in U.S. Pat. No. 5,029,509 entitled, “Musical synthesizer combining deterministic and stochastic waveforms” by X. Serra and J. Smith. To begin, a sound signal
102
is recorded from an acoustic microphone. In step
104
, a time frame is set and a fundamental pitch is determined for the sound signal
102
, in step
106
. Next in step
108
, one or more sine wave functions are generated with the same time frame and harmonic relationships as the sound signal
102
. The sine functions are fitted, in step
109
, to the sound signal
102
and, in step
110
, a corresponding set of amplitudes, phases, and frequencies for the sine functions are stored in memory
112
. The sine functions, best fit by one or more harmonic waves, are then subtracted from the sound signal
102
to generate a residual acoustic signal
114
. The residual signal
114
is then fitted
116
to a white noise signal from a white noise generator
118
. The white noise signal can be further characterized using a Linear Predictive Coding (LPC) technique, also called an all-pole method technique. Next in step
120
, coefficients that fit the noise signal to the residual signal
114
are stored in the memory
112
and are recalled on command.
As a first step in synthesizing the sound signal
102
, MIDI or other sound control systems call for the needed sequence of amplitude, pitch, attack, reverberation, and other control variables
122
. Next, in step
126
, the coefficients stored in the memory in step
120
that describe the sine and residuals are modified to generate variations of the stored sounds, according to the user's control description. In step
128
, sequences of sine functions describing the harmonic aspects of the desired synthesized signal are added together. The functions describing the residual signal
114
are added to the summed harmonic functions to make a single time frame of a sound sequence signal
130
. Finally, sequential frames are added together to make a multiframe sound sequence signal
130
, which is then amplified through a speaker
132
for a listener
134
to hear. Ideally the multiframe sound sequence signal
130
is as close a match as possible to the original sound signal
102
.
A variant on the third prior art method is described in U.S. Pat. No. 5,587,548 entitled, “Musical Tone Synthesis System Having Shortened Excitation Table,” by Smith. Smith describes how to use previously measured sound excitations, usually obtained by impulsive excitation of body modes, to synthesize a musical sound. In addition, Smith uses computed “loops” whose cycle rate defines pitch values for synthesized sound. This process is called “physical modeling synthesis.” This process often uses simplified functional descriptions of vibrating mechanical elements to describe the mechanics of string motions, their coupling to the body via a bridge, and to resonator panels in musical instruments.
None of these prior art methods accurately captures qualities of inanimate sound sources because they are only based upon approximating an output signal, or intermediate process, and fail to accurately characterize the underlying physical processes of sound generation and their collective behavior. Such methods also have difficulty defining time frames based upon a natural cycle of the sound sources, and in forming accurate transfer functions and associated filters, especially in high noise environment. As a result, prior art methods are not able to accurately synthesize well-known musical instruments, such as Steinway pianos, Stradivarius violins, ebony clarinets, and other fine instruments to the degree desired. In addition, these prior art analysis methods tend to not work well in real time sound cancellation applications.
Separate from the three methods described above, hybrid sound generators, such as are used in electric guitars, generate sounds by monitoring excitation sources, such as strings on a musical instrument. In these hybrid instruments, acoustic sounds from the musical instrument itself are usually ignored. U.S. Pat. No. 5,572,791 entitled, “Method for positioning a pickup on an electric guitar” by K. Kazushige and U.S. Pat. No. 4,321,852 entitled, “Stringed instrument synthesizer apparatus” by L. D. Young Jr. describes these techniques and methods in more detail. As a result, typical guitar string sensors only measure approximate excitations of the musical instrument, which are then fed to an amplifier, filter bank, and loud speaker.
The prior art methods of sound analysis and synthesis are also applied to the problem of canceling out undesirable sounds. These methods are described in references such as the
Encyclopedia of Acoustics
, Vols. I-IV, M. J. Crocker ed., Wiley, N.Y. 1997; Vol. II chapters 79-89 and Vol. IV chapters 130-139, in U.S. Pat. No. 5,517,571 entitled, “Active noise attenuating device of the adaptive control type” by S. Saruta & Y. Sekiguchi, and also in U.S. Pat. No. 5,448,645 entitled, “Active fan blade noise cancellation system” by J. R. Guerci. Each of these sound cancellation methods, however, is incapable of rapidly and economically analyzing and canceling out the undesirable sounds to the degree needed, especially for rapidly changing (e.g., chaotic) sounds.
In practice, obtaining accurate mathematical functions of complex sound systems is currently very difficult. Prior art excitation transducers are not accurate or fast enough and prior art algorithms are restricted to using LPC methods. Furthermore, digital processors are too slow and expensive to handle needed information, and required memories are costly. As a result, automating sound analysis so that a wide variation in sound inputs can be automatically analyzed and stored in memory for subsequent synthesis, especially in “real time,” is very difficult, if not impossible in some cases. There is a need in the art for more accurate systems, for faster systems, and methods for more accurately characterizing, synthesizing, and/or canceling out acoustic signals from
Burnett Greg C.
Holzrichter John F.
Ng Lawrence C.
Broda Samuel
Dakin Lloyd
Phan T.
Scott Eddie E.
The Regents of the University of California
LandOfFree
System and method for characterizing synthesizing and/or... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method for characterizing synthesizing and/or..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for characterizing synthesizing and/or... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3015744