Apple Computer Inc. v. Burst.com, Inc.

Filing 109

Declaration of Sheila Hemami in Support of 107 Response Burst.com, Inc's Opposition to Plaintiff Apple Computer, Inc.'s Motion for Summary Judgment on Invalidity Based on Kramer and Kepley Patents filed byBurst.com, Inc.. (Attachments: # 1 Exhibit A to S. Hemami Declaration# 2 Exhibit B to S. Hemami Declaration# 3 Exhibit C to S. Hemami Declaration# 4 Exhibit D to S. Hemami Declaration# 5 Exhibit E to S. Hemami Declaration# 6 Exhibit F to S. Hemami Declaration# 7 Exhibit G to S. Hemami Declaration# 8 Exhibit H to S. Hemami Declaration# 9 Exhibit I to S. Hemami Declaration# 10 Exhibit J to S. Hemami Declaration# 11 Exhibit K to S. Hemami Declaration)(Related document(s) 107 ) (Crosby, Ian) (Filed on 6/7/2007)

Apple Computer Inc. v. Burst.com, Inc.

Doc. 109 Att. 7

Case 3:06-cv-00019-MHP

Document 109-8

Filed 06/07/2007

Page 1 of 5

Dockets.Justia.com

Case 3:06-cv-00019-MHP

Document 109-8

Filed 06/07/2007

Page 2 of 5

SPEECH/SILENCE SEGMENTATION FOR REAL-TIME CODING VIA RULE BASED ADAPTIVE ENDPOINT DETECTION
J. F. Lynch Jr. J. 6. Josenhans R. E. Crochiere
Speech Processing Department AT&T Bell Laboratories Murray Hill, New Jersey 07974

ABSTRACT: A new algorithmic technique is presented for efficiently implementing the end-point decisions necessary to separate and segment speechfromnoisy background environments. The algorithm utilizes a set of computationally efficient production rules that are used to generate speech and noise metrics continuously from the input speech waveform. These production rules are based on statistical assumptions about the characteristics of the speech and noise waveform and are generated via time-domain processingtoachieve a zero delay decision. Anend-pointercompares the speech and silence metrics using an adaptive thresholding scheme with a hysteresis characteristic to control the switching speed of the speech/silence decision. The paper further describes the application of thisalgorithm to silencecompression of speechinwhichspeech segments are encoded with a low bit-rate encoding scheme and silence information is characterized by aset of parameters. In the receiver the resulting packetized speech is reconstructed by decoding the speech segments andreconstructing the silence intervalsthrougha noise substitution process in which the amplitudeandduration of backgroundnoiseis defined by the silence parameters. A noise generation technique is described which utilizes an order 18th polynomial to generate a spectrally flatpseudo-randomsequence that is filtered to match the mean coloration of acoustical background noise. A technique is further describedinwhich the speech/silence transitions are merged rather than switched to achieve maximum subjective performance of the compression technique. The above silence compression algorithm has been implemented in a single DSP-20 signal processing using chip sub-band coding for speech encoding. Using this system, experimentswere conducted to evaluate the performance of the technique and to verify the robustness of the endpoint and silence compression over a wide range of background noise conditions.

PI. Speech and Noise Metric Generation
Silence compression uses heuristic production rules to generate speech and noise metrics continuouslyfrom the input signal. These metrics are used later to generate the causal speech/noisedecision. The first stepto designing these production rules is to state some known properties about the speech and noise signals provide to a foundation for the design. The following statements give a few properties of the speech signal that will be put touse later in the production rules: 1. Empirical evidence shows segments contain "talk-spurts" duration [11. that 99.9% of "continuous speech" in of less than 2.0 seconds

2. Empirical evidence shows also that 99.56% of "continuous speech" segments have gaps of less than 150 ms. in duration [ 11.
3. Speech energy can

only increase background acoustic level.

the signal level above

the

Statements about the expected nature of the "background acoustic noise" are more difficult to make since it is implicitly undefined. However in apractical sense the definition can benarrowed. Here, backgroundnoiseisassumed to be the relatively stationary "hum" or "buzz" due to ventilating systems, computer equipment or the aggregate background sounds of an office environment. Wonstationary background sounds such as nearby conversations among workers and nearby ringing telephones are classed as speech sounds and not background noise. The following statements can be made about background noise in this context: Background acoustic noiseenergydecreases approximately 5 dB per octave [21. A 60 Hz highpassfilter sonic background noise. is usedin with frequency at

the system to remove sub-

I. Introduction
This