Speech activated control system and related methods

Patent No. US7774202 (titled "Speech activated control system and related methods") on Jun 12, 2006. The application was issued on Aug 10, 2010.

How does this patent fit in bigger picture?

Technical Landscape

In the mid-2000s when ’202 was filed, speech recognition systems were typically implemented using static vocabulary databases that required extensive pre-training and fixed syntax structures. At a time when systems commonly relied on pre-determined command sets embedded within the speech engine, hardware and software constraints made real-time, in-flight adaptation to changing operational environments non-trivial. Engineering practices of this era often struggled with high-noise environments, such as aircraft cockpits, where background interference frequently degraded recognition accuracy and necessitated manual user confirmation of commands, thereby increasing operator workload.

Prosecution Position

The disclosed invention represents a technical advancement through an architectural shift that divorces the speech recognition engine from specific vehicle command functionality via a dedicated data interface. This integration enables a real-time, user-programmable control environment where operators can define custom switch macros and associate them with arbitrary vocal annunciations during operation. The system achieves enhanced robustness in high-noise environments by implementing a nonlinear noise removal process that thresholds equalized spectral bins to isolate speech formants. Furthermore, the capability to generate language-independent models based on sound components rather than fixed words allows for dynamic template creation that is adaptable to different users and operational profiles without reconfiguring the underlying engine.

Claims

This patent contains 31 claims, with claims 1, 12, and 27 serving as the independent claims. The independent claims focus on a speech-activated control system and related methods for aerial vehicle components, specifically detailing a noise reduction and speech enhancement process that involves dividing speech into time slices and frequency bins, estimating noise power, equalizing energy values, and thresholding to isolate speech formants for model generation or recognition. The dependent claims further define the invention by specifying real-time in-flight execution, confidence scoring, switch macro recording, signal integrity checks such as clipping and cropping analysis, specific alignment techniques, and the use of various speech recognition engines like Hidden Markov Models and Neural Networks.

Key Claim Terms New

Definitions of key terms used in the patent claims.

Term (Source)Support for SpecificationInterpretation
Feature vectors
(Claim 1, Claim 12, Claim 27)
A parameterized feature extraction process is employed in which a Short-Time Fourier analysis is used to produce either an associated spectrum or a cepstrum. The cepstrum involves applying additional nonlinear processing to the spectrum and another transform back to a quasi-time domain feature space. The instructions include developing a set of feature vectors representing energy of a frequency content of the digitized user-speech template to determine a unique pattern identifying the command annunciation.A set of mathematical representations of the energy and frequency content of a speech signal used to identify unique patterns for recognition.
Frequency range-specific noise power estimates
(Claim 1, Claim 12, Claim 27)
Background noise estimates are acquired in each spectral bin. The background noise contour for noise within the observation frame or window can be determined and the noise can be removed from within and around speech formants of the sampled data. This is accomplished by first estimating noise power for each spectral bin to define a background noise contour.Calculated values representing the estimated power of background noise for specific frequency bands (bins) across multiple time segments of a speech signal.
Integrity check
(Claim 12)
Data representing received speech or a command annunciation is subjected to an integrity check and is pre-aligned using a convolution-based algorithm. Post-feature processing can be applied to the signal in which cropping and/or clipping effects can be identified and/or corrected. Severe cropping or clipping, especially during training, are to be detected so that the utterance is prevented from being applied to form recognition models to ensure integrity.A diagnostic process performed on sampled audio data to detect quality issues such as clipping, cropping, low volume, or misalignment that would degrade recognition accuracy.
Observation window
(Claim 12)
The speech removal process can employ a novel convolution-based algorithm to align isolated speech at a central position of an observation frame. According to an embodiment, the size of the utterance observation window encapsulating the speech can be configured to meet the operational requirements. The background noise contour for noise within the observation frame or window can be determined.A configurable temporal frame or sampling interval that encapsulates the speech signal for alignment and analysis.
Speech formants
(Claim 1, Claim 12, Claim 27)
The background noise contour is determined and normalized to uniformity, and bins determined to contain noise are then removed from within and around the speech formants. The speech recognizer can first perform noise removal and equalization where signals are thresholded based on noise estimates and the spectrum is adjusted based on acquired response characteristics. This process removes noise from within and around speech formants of the sampled data using a nonlinear noise removal process.The spectral peaks or resonant frequencies of the human voice within a speech signal that carry the primary information of the utterance.

Patent Summary

Litigation Cases New

US Latest litigation cases involving this patent.

Case NumberFiling DateTitle
4:25-cv-00503May 22, 2025Simplehuman Llc V. Hms Manufacturing Company Llc

Patent Family

Patent Family

File Wrapper

The dossier documents provide a comprehensive record of the patent's prosecution history - including filings, correspondence, and decisions made by patent offices - and are crucial for understanding the patent's legal journey and any challenges it may have faced during examination.

  • Get instant alerts for new documents

US7774202

Application Number
US11451217A
Filing Date
Jun 12, 2006
Publication Date
Aug 10, 2010
External Links
Slate, USPTO , Google Patents