Obviousness

Obviousness Analysis of US Patent 10,306,389 Under 35 U.S.C. § 103

This analysis assesses the obviousness of US Patent 10,306,389 (the '389 Patent) under 35 U.S.C. § 103, considering prior art references and the perspective of a person having ordinary skill in the art (POSITA) as of the patent's priority date of March 13, 2013.

The '389 Patent broadly describes a head-wearable acoustic system with noise-canceling microphone geometry, apparatuses, and methods. Key inventive concepts include:

1. Microphone Geometry for Signal-to-Noise Ratio (SNR) Difference: Employing main and reference microphones positioned on a head-wearable device to achieve a superior SNR for desired audio at the main microphone (SNR_M) compared to the reference microphone (SNR_R). This difference is attributed to varying acoustic path lengths and/or the use of different directivity patterns or beamforming [cite: "the invention", FIG. 1, FIG. 2, FIG. 6, FIG. 7].
2. Multi-Stage Noise Cancellation: A cascaded approach combining multi-channel adaptive noise cancellation (ANC) with single-channel linear noise cancellation (SCNC) to effectively remove undesired audio [cite: "noise cancellation architectures", FIG. 21, FIG. 24B].
3. Linear Signal Processing: Maintaining linearity in signal processing to preserve the quality and fidelity of desired audio, which benefits speech recognition (SR) and automatic speech recognition (ASR) algorithms [cite: "linear signal processing", FIG. 22].
4. Desired Voice Activity Detection (VAD) with Multi-Channel Compression: Utilizing acoustic signals from multiple channels with signal compression (e.g., logarithmic) to generate a normalized main signal for robust voice activity detection [cite: "multi-channel acoustic signal compression", FIG. 28A].
5. Auto-Balancing: Employing an auto-balancing unit to compensate for microphone response drift over time, thereby maintaining system performance [cite: "acoustic channels", FIG. 24A, FIG. 29A].
6. Integration into Head Wearable Devices: Implementing these systems within various head-wearable devices such as glasses, goggles, visors, and helmets [cite: FIG. 3A, FIG. 8-19].

Identified Prior Art References

From the provided patent text, United States Patent 7,386,135 (Ratzel et al.), titled "Cardioid Beam With A Desired Null Based Acoustic Devices, Systems and Methods," is explicitly incorporated by reference [cite: "U.S. Pat. No. 7,386,135"]. Ratzel et al. has a priority date of October 4, 2004, and an issue date of June 10, 2008, making it effective prior art against the '389 Patent.

Obviousness Combinations and Motivation

A person having ordinary skill in the art (POSITA) in the field of acoustic signal processing for wearable devices, as of March 2013, would have been motivated to combine the teachings of Ratzel et al. with general knowledge and other known techniques to arrive at the claimed invention.

Combination 1: Ratzel et al. (US7386135B2) + General Knowledge of Wearable Acoustics + Standard Noise Cancellation Architectures

Ratzel et al. (US7386135B2): This patent teaches the fundamental concept of using multiple microphones to create directional acoustic beams with nulls, specifically a cardioid beam, for noise cancellation in acoustic devices [cite: "the implementation and operation of the main channel activity detector 2206, the reference channel activity detector 2208 and the inhibit logic 2214"]. This directly addresses creating an SNR difference between channels by exploiting directivity patterns, where a null can be directed towards undesired noise or away from desired speech for a reference microphone. Ratzel also describes multi-channel voice activity detection and associated control logic.

Motivation for Combination:

1. Placement on a Head-Wearable Device for SNR Difference: A POSITA would be well-aware of the increasing demand for hands-free communication and interaction through wearable devices, especially head-worn form factors (e.g., Bluetooth headsets, smart glasses prototypes). Given Ratzel et al.'s teaching of using microphone directivity to enhance desired speech and suppress noise, it would be an obvious design choice to adapt this principle to a head-wearable device. Placing a "main" microphone closer to the user's mouth and a "reference" microphone further away, or orienting them with different directivity patterns (e.g., one pointing towards the mouth, another with a null towards the mouth as taught by Ratzel [cite: "FIG. 7 illustrates, generally at 700, a misaligned reference microphone response axis according to embodiments of the invention"]), would inherently create the desired SNR difference between desired speech and background noise. The acoustic principle that sound attenuates with distance (spherical spreading) and that the human head causes shadowing effects were well-known and would naturally lead to different acoustic path lengths and signal amplitudes at microphones placed at different locations on a user's head, thus creating an SNR difference.
2. Multi-Stage Noise Cancellation: By 2013, the field of audio signal processing had well-established techniques for noise reduction. Adaptive Noise Cancellation (ANC) using multiple microphones was known to be effective for removing correlated noise and reverberation. Single-Channel Noise Cancellation (SCNC) methods, such as spectral subtraction or Wiener filtering, were also common for reducing stationary or residual noise. A POSITA seeking to achieve robust noise cancellation in diverse and challenging acoustic environments (e.g., vehicle cabins, public spaces, as mentioned in the '389 Patent [cite: "Stationary and non-stationary sources of undesired audio exist in office environments, concert halls, football stadiums, airplane cabins, everywhere that a user will go with an acoustic system"]) would be motivated to combine these complementary techniques in a cascaded manner. An adaptive filter handles initial broad-spectrum noise reduction using multiple inputs, followed by a single-channel filter to refine the output by addressing remaining stationary noise components, as shown in FIG. 21 of the '389 Patent. This is a common engineering strategy to leverage the strengths of different algorithms.
3. Linear Signal Processing for Speech Quality: The importance of maintaining linear signal processing to preserve speech quality and avoid distortion that could degrade the performance of subsequent speech recognition (SR/ASR) algorithms was a known principle in audio engineering. Thus, a POSITA would inherently aim for linear processing in a noise cancellation system intended for communication or voice command applications.
4. Desired Voice Activity Detection (VAD) with Multi-Channel Compression: Ratzel et al. already disclose multi-channel voice activity detection and associated inhibit logic for filter control. A POSITA, seeking to improve the robustness of VAD in noisy environments, would find it obvious to apply known signal processing techniques like compression (e.g., logarithmic compression to handle wide dynamic ranges more effectively) to the power calculations from the main and reference channels. This would lead to a more stable and effective "normalized main signal" for threshold comparison, as described in FIG. 28A of the '389 Patent, which could then be used to control the adaptive and single-channel filters.

Combination 2: Ratzel et al. (US7386135B2) + General Knowledge of System Reliability and Calibration

Ratzel et al. (US7386135B2): Provides a foundation for multi-microphone acoustic systems.

Motivation for Combination:

1. Auto-Balancing: In any multi-microphone system, particularly those operating in real-world conditions over extended periods, component variations and environmental factors (temperature, humidity, aging) can cause microphone sensitivities and channel gains to drift. This drift directly degrades the performance of noise cancellation algorithms that rely on precise relationships between microphone signals. A POSITA designing a commercial product for consistent long-term performance would be highly motivated to incorporate mechanisms for maintaining channel balance. Implementing an "auto-balancing unit" (as depicted in FIG. 24A and FIG. 29A of the '389 Patent) to periodically or continuously calibrate and adjust the gains of the microphone channels would be an obvious engineering solution using known calibration techniques (e.g., by analyzing far-field noise during periods of no desired speech or injecting known test signals) to prevent performance degradation due to sensitivity drift.

Conclusion of Obviousness

The '389 Patent's claims, interpreted broadly from the provided description, appear to be an obvious combination of known acoustic and signal processing principles applied to a head-wearable device. The core concept of using microphone geometry (placement and/or directivity) to create an SNR difference between main and reference channels for noise cancellation is strongly suggested by Ratzel et al. (US7386135B2). The further refinements, such as cascading adaptive and single-channel noise cancellation, employing multi-channel VAD with compression, and implementing auto-balancing, represent logical and well-known engineering choices for improving the performance and reliability of multi-microphone acoustic systems, particularly for commercial wearable applications. A POSITA would have been motivated to combine these known elements to create a more effective and robust noise-canceling system for head-wearable devices.