20240020526. SYSTEMS AND METHODS FOR FALSE POSITIVE MITIGATION IN IMPULSIVE SOUND DETECTORS simplified abstract (Robert Bosch GmbH)

From WikiPatents
Jump to navigation Jump to search

SYSTEMS AND METHODS FOR FALSE POSITIVE MITIGATION IN IMPULSIVE SOUND DETECTORS

Organization Name

Robert Bosch GmbH

Inventor(s)

Luca Bondi of Pittsburgh PA (US)

Samarjit Das of Wexford PA (US)

Shabnam Ghaffarzadegan of Livermore CA (US)

SYSTEMS AND METHODS FOR FALSE POSITIVE MITIGATION IN IMPULSIVE SOUND DETECTORS - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240020526 titled 'SYSTEMS AND METHODS FOR FALSE POSITIVE MITIGATION IN IMPULSIVE SOUND DETECTORS

Simplified Explanation

The abstract describes a method for data augmentation in audio processing. The method involves receiving audio stream data associated with impulse events and their labels. It then detects the peaks of the impulse events using an onset detector. The method extracts positive samples of the audio stream data associated with the impulse events and applies the corresponding labels. It also extracts negative samples of the audio stream data associated with the impulse events. The method augments training data by using the positive and negative samples and trains machine-learning models using the augmented data.

  • The method receives audio stream data and labels associated with impulse events.
  • It detects the peaks of the impulse events using an onset detector.
  • Positive samples of the audio stream data associated with the impulse events are extracted.
  • The labels are applied to the positive samples.
  • Negative samples of the audio stream data associated with the impulse events are extracted.
  • Training data is augmented using the positive and negative samples.
  • At least one machine-learning model is trained using the augmented training data.

Potential Applications:

  • Audio event detection and classification
  • Speech recognition and transcription
  • Environmental sound analysis
  • Music information retrieval

Problems Solved:

  • Insufficient training data for machine-learning models
  • Difficulty in detecting and classifying impulse events in audio streams

Benefits:

  • Improved accuracy and performance of machine-learning models
  • Enhanced audio analysis and understanding capabilities
  • Increased efficiency in audio processing tasks


Original Abstract Submitted

a method data augmentation includes receiving audio stream data associated with at least one impulse event, receiving a label associated with the audio stream data, and detecting, using an onset detector, at least one peak of the at least one impulse event. the method also includes extracting at least one positive sample of the audio stream data associated with the at least one impulse event. the method also includes applying, to the at least one positive sample, the label associated with the audio stream data and extracting at least one negative sample of the audio stream data associated with the at least one impulse event. the method also includes augmenting training data based on the at least one positive sample and the at least one negative sample and training at least one machine-learning model using the augmented training data.

Retrieved from "http://wikipatents.org/index.php?title=20240020526._SYSTEMS_AND_METHODS_FOR_FALSE_POSITIVE_MITIGATION_IN_IMPULSIVE_SOUND_DETECTORS_simplified_abstract_(Robert_Bosch_GmbH)&oldid=28221"