Data processing: speech signal processing – linguistics – language – Speech signal processing – Application
Reexamination Certificate
1998-06-11
2001-05-08
Hudspeth, David (Department: 2741)
Data processing: speech signal processing, linguistics, language
Speech signal processing
Application
C084S662000
Reexamination Certificate
active
06230140
ABSTRACT:
©Copyright Frederick Severson and Patrick Quinn, 1993: The disclosure of this patent document contains material that is subject to copyright protection. The copyright owners have no objection to facsimile reproduction by anyone of the patent document or the patent disclosure, as is appears in the Patent and Trademark Office patent file or records, but otherwise reserve all copyrights whatsoever.
FIELD OF THE INVENTION
The present application also pertains to the field of electronic sound effects. In particular, this invention describes new ways to generate continuous digital sound from a limited number of stored sound records using random and/or logical selection techniques. This non-looped method keeps the sounds constantly changing in a realistic manner for applications requiring continuous sound effects such as background seashore sounds, cracking fireplace sounds, drums, etc.
BACKGROUND OF THE INVENTION
Creating continuous sounds using looping techniques has been around since tape recorders were available commercially. The idea with a looped sound is to play out a sound record over and over again in a continuous loop to produce a sound of indefinite length from a single short sound record. For instance, the sound of a drum as a background beat can be created by having a single drum-hit sound record played over and over. This is useful in musical synthesizers where a selected drum record can be played out in a loop to provide a constant drum beat to set the timing of the music. This eliminates the need to have a entire drum sequence stored in digital memory or the need for the operator to constantly strike a drum key to maintain the beat.
The problem with any looped sound is that it will get quite boring or irritating when listened to for long periods of time. The mind quickly detects a pattern when the same sound is played over and over and starts to expect or anticipate the next loop segment. The effect is for the mind to either try to tune the sound out or to be distracted by it like listening to the sound of dripping faucet.
Sounds in Nature never precisely repeat. Even when a sound appears to be repetitive, there are always slight differences that tell the listener that the sounds are being created new each time. A drum that is hit consistently, sounds a little different each time and each revolution of a motor that seems to be generating a constant sound, has its own variability. It is this variability that is missing in looped sound that clues the listener in to the fact that it is the same sound record repeated over and over and tells the listener that it is not a realistic sound effect. Most sounds in Nature or sounds made by man-made machines are not truly periodic. There are slight changes in amplitude and phase that make each “apparent period” slightly different in content. For instance, consider the case of man-made motor sounds of a constant running diesel engine. No matter how smooth a motor is running, it can not run perfectly. There are always slight changes in amplitude or misses or RPM changes that make motors sound like motors. It is the combination of randomness and repetitiveness that gives sounds of machines a “real quality”. You know what a motor is going to sound like from second to second—but not quite. You don't know if it will “miss” or speed up slightly or whatever. But what you don't expect is for it to sound perfectly the same.
If we make up motor sounds or horns or any so called “repetitive sound” by looping a short sound record, the sound would be too perfect, too exact and too boring to be believable.
Other continuous sounds that are not considered periodic do even worse when played in a continuous loop. For instance, the sound of children playing could be recorded for some finite time (say 10 seconds) and then played back in a continuous loop to model background playground sounds. However, if there are any distinctive sounds, like a child yelling a phrase or something, the listener will quickly recognize the sound record and the effect will become expected. The sound will have lost its innocent quality and will become irritating. The sound of babbling brooks, wood fires, water falls, traffic sounds, sea shore waves, etc. are all continuous non-repeating sounds that are poorly modeled by a repetitive loop.
SUMMARY OF THE INVENTION
Non-looped Continuous Sound by Random Sequencing of Digital Sound Records can be abbreviated in name to Random Sequenced Sound, and further abbreviated as RSS. RSS, at its core, consists of taking several short segments of an otherwise continuous sound, and making independent records of each of these short segments. Then these independent segments are re-assembled into a continuous, never-repeating sound sequence based on selecting the next sound segment according to some statistical algorithm. This statistical algorithm itself may be chosen by various circumstances (such as the passage of time, or the coincidence with some other sound effect, changes in ambient light, heat, operator input of some sort, etc.) or perhaps selected from a library of algorithms in a deterministic or random way. Also, there are RSS situations where you will want to switch from one set of sound segments to an alternate set of sound segments. Examples might be birds in the morning and crickets at night. Or, perhaps, crickets until a model train comes by and then barking and howling dogs. There may or may not be specific silent pauses inserted between each record, depending on the desired effect. Also, the sound segments may be sequenced in a logical but statistical way depending on the content of the previous sound record or other inputs to the system.
Random sequenced sound is a way to produce continuous sounds that follow a theme and vary in content by logical and/or statistical methods in order to model the continuity, variability, and logical progression of sounds heard in real life. It is not simply a method to play different sound records in a random way such as might be produced from a CD machine playing different songs in a random order.
Accordingly, one aspect of the invention is a method of generating random sequenced sound. In general, the method includes the steps of storing a sound record; dividing the sound record into a plurality of sound segments; providing a desired probability density finction “pdf”; selecting one of the sound segments according to the probability density function; playing the selected sound segment; and repeating said selecting and playing steps thereby generating non-looped continuous sound.
The probability density function pdf can take many forms. For example, it may provide a uniform distribution so that each of the sound segments is equally likely to be selected and played at any time. In another embodiment, the probability density function may provide an approximately Gaussian distribution so that certain sound segments are more likely to be selected and played than the other sound segments. Any desired “weighting” of the sound segments can be implemented by the pdf.
“Logical Sequenced Sound” (LSS) features selection of sound segments not merely in random sequence, but in a logical sequence whereby each selection depends on one or more selections that preceded it. Alternatively, LSS may be used for selecting an appropriate group of sound segments, while the selection and playing of individual segments within the group proceeds according to a predetermined function such as a pseudo-random selection.
This invention describes methods to produce continuous sounds that do not have a recognizable pattern which, in turn, keeps the sounds fresh and non-predicable and reduces the irritation common with looped sound effects. This invention has application in many areas. Some of these are listed below:
background sound effects for model layouts such as model railroading
soothing, restful, or interesting background sound for homes or businesses
producing sustained sound such as the middle portion of a digitally produced horn
in music synthesizer rhythm generation
sound effects to deter burglary
sou
Quinn Patrick A.
Severson Frederick E.
Hudspeth David
Lerner Martin
Stoel Rives LLP
LandOfFree
Continuous sound by concatenating selected digital sound... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Continuous sound by concatenating selected digital sound..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Continuous sound by concatenating selected digital sound... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2496294