Obviousness

Obviousness Analysis of US Patent 8,320,575 under 35 U.S.C. § 103

This analysis examines the obviousness of the independent claims of US Patent 8,320,575 in light of prior art available before its priority date of October 1, 2007. A claim is considered obvious if the differences between the claimed invention and the prior art are such that the subject matter as a whole would have been obvious to a Person Having Ordinary Skill in the Art (PHOSITA).

A PHOSITA in the field of audio signal processing circa 2007 would typically have a master's degree in electrical engineering or a related field, along with several years of experience in digital signal processing (DSP), particularly in audio applications like echo cancellation and noise suppression. Such a person would be familiar with frequency-domain processing techniques like the Fast Fourier Transform (FFT), filter banks, and the trade-offs between computational complexity and signal quality.

---

Analysis of Independent Claim 1

Claim 1: A method for audio signal processing, comprising:

1. dividing an audio signal into audio sub-band signals;
2. excising a subset of the audio sub-band signals;
3. processing a remaining subset of the audio sub-band signals to obtain enhanced audio sub-band signals;
4. reconstructing at least a portion of the subset of the audio sub-band signals that were excised; and
5. synthesizing the enhanced audio sub-band signals with the reconstructed audio sub-band signals to form an enhanced audio signal.

Proposed Obviousness Combination:
The teachings of US5272695A (hereafter '695) in view of US20070071277A1 (hereafter '277) and the general knowledge of a PHOSITA regarding computational efficiency would have rendered Claim 1 obvious.

Reasoning:

1. Motivation to Combine:
   A PHOSITA is constantly driven to reduce the computational complexity and memory requirements of DSP algorithms to enable their use in real-time, low-power applications like hands-free car kits or mobile phones. The method of processing signals in the sub-band (frequency) domain was a well-established technique for this purpose. The '695 patent teaches a sub-band echo canceller, a computationally intensive application. A PHOSITA, when implementing such a system, would be motivated to find further efficiencies. It would have been obvious to consider processing only a subset of the sub-bands if doing so would significantly reduce the computational load without unacceptably degrading the audio quality. The idea of discarding less critical information to improve processing speed is a fundamental trade-off in computer science and signal processing.

2. Mapping of Prior Art to Claim Elements:

   • "dividing an audio signal into audio sub-band signals": This is explicitly taught by '695 in the context of echo cancellation. The abstract of '695 states its method involves a "subband echo canceller" which inherently requires dividing the signal into sub-bands. This is a foundational step in sub-band processing.

   • "excising a subset of the audio sub-band signals" and "processing a remaining subset": The '695 patent teaches processing the sub-band signals to perform echo cancellation. While it does not explicitly teach excising sub-bands to be reconstructed later, the motivation to do so is high for a PHOSITA seeking efficiency. The '277 patent application, in the context of watermarking, teaches modifying only selected sub-bands. For instance, '277 states, "For each sub-band, a decision is made whether to embed a watermark bit or not" (para.). This demonstrates the principle of selectively processing a subset of sub-bands while leaving others unprocessed or processed differently. A PHOSITA would readily understand that forgoing processing on some bands (i.e., "excising" them from the main processing path) would save MIPS (Millions of Instructions Per Second) and memory.

   • "reconstructing at least a portion of the subset of the audio sub-band signals that were excised": The concept of reconstructing or interpolating missing data is a standard technique in signal processing. The '277 patent application teaches the synthesis of a full-band signal from modified and unmodified sub-bands, which is analogous to reconstruction. More specifically, a PHOSITA would know that if, for example, every other frequency bin (sub-band) were discarded, the missing bins could be estimated by interpolating between the adjacent, processed bins. This is a common method for data reconstruction. The '575 patent itself describes this as a form of interpolation (Column 8, lines 16-24). Such interpolation techniques were well within the common general knowledge of a PHOSITA at the time.

   • "synthesizing the enhanced... and reconstructed... signals to form an enhanced audio signal": This final step is the necessary conclusion to any sub-band processing scheme. Both '695 and '277 teach the use of a synthesis filter bank to combine the processed sub-bands back into a single time-domain signal. It is the inverse operation of the initial "dividing" step.

Therefore, a PHOSITA starting with a standard sub-band processing system like that in '695 would have been motivated by the perpetual need for efficiency to selectively process only a subset of the sub-bands (as suggested by the principle in '277) and then reconstruct the unprocessed bands using standard interpolation techniques, thus arriving at the method of Claim 1.

---

Analysis of Independent Claim 10

Claim 10: A method for audio signal processing, comprising the steps of Claim 1, wherein the act of processing comprises:

• dividing a reference signal into reference sub-band signals;
• excising a subset of the reference sub-band signals that is equal in number to the excised subset of the audio sub-band signals;
• adapting filter coefficients of an echo compensation filter based on a remaining subset of the reference sub-band signals; and
• using the adapted filter coefficients to remove echo components from at least a portion of the remaining subset of the audio sub-band signals.

Proposed Obviousness Combination:
The teachings of US5272695A ('695) alone or in view of the same efficiency motivations discussed for Claim 1 would have rendered Claim 10 obvious.

Reasoning:

The '695 patent is titled "Subband echo canceller." The entire purpose of the patent is to describe an apparatus and method for echo cancellation in the sub-band domain. An echo canceller, by its very definition, operates by:

1. Receiving a primary audio signal (e.g., from a microphone) that contains an echo.
2. Receiving a "reference signal" (e.g., the far-end speaker signal being played out of a local loudspeaker) which is the source of the echo.
3. Using an adaptive filter to model the echo path and subtract an estimated echo from the primary signal.

The '695 patent explicitly teaches these steps in the sub-band domain.

1. Mapping of Prior Art to Claim Elements:

   • "dividing a reference signal into reference sub-band signals": This is inherent and necessary for the sub-band echo canceller in '695. To perform echo cancellation in the sub-band domain, both the microphone signal and the reference loudspeaker signal must be transformed into the same sub-bands to be compared and processed.

   • "excising a subset of the reference sub-band signals that is equal in number to the excised subset of the audio sub-band signals": This is an obvious and necessary consequence of the primary inventive concept of excising sub-bands for efficiency. If the processor is not going to enhance a particular sub-band of the microphone signal (because it has been "excised"), then there is no reason to process the corresponding sub-band of the reference signal for that purpose. A PHOSITA would immediately understand that to maintain correspondence, the same sub-bands must be ignored for both signals. Processing reference sub-bands that correspond to excised audio sub-bands would be useless and would defeat the entire purpose of the efficiency gain.

   • "adapting filter coefficients of an echo compensation filter based on a remaining subset of the reference sub-band signals" and "using the adapted filter coefficients to remove echo components from ... the remaining subset of the audio sub-band signals": This is the central teaching of '695. The patent describes adapting the filter coefficients for each sub-band based on the reference signal for that sub-band, and then using that filter to cancel the echo. Applying this known method only to the "remaining" (non-excised) sub-bands is a direct and obvious application of the efficiency concept.

In conclusion, Claim 10 merely recites a known sub-band echo cancellation architecture ('695) and applies to it the obvious efficiency-improving technique of skipping the processing for a subset of the sub-bands and reconstructing them later. The additional steps recited in Claim 10 are simply the necessary and logical consequences of applying the method of Claim 1 to the specific, well-known application of sub-band echo cancellation.