Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

Perceiving Loudness, Pitch, and Location01:21

Perceiving Loudness, Pitch, and Location

641
The human brain perceives pitch through two primary mechanisms reflected in place theory and frequency theory. Each mechanism describes how sound waves are interpreted as specific pitches by the brain, offering insights into the intricate processes of auditory perception.
Place theory, or place coding, suggests that different pitches are heard because various sound waves activate specific locations along the cochlea's basilar membrane. The brain determines the pitch of a sound by...
641
Difference from Background: Limit of Detection01:05

Difference from Background: Limit of Detection

7.7K
The limit of detection (LOD) is the smallest amount of analyte that can be distinguished from the background noise. The LOD value corresponds to the concentration at which the analyte signal is three times larger than the standard deviation of the blank signal. Below this value, the analyte signal cannot be differentiated from the background noise. It is calculated by dividing the calibration slope by 3 times the standard deviation of the blank signals.
The LOD indicates the presence or absence...
7.7K

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

For better and worse: neural self-partner overlap during social feedback is associated with relationship satisfaction and depressive symptoms.

Cerebral cortex (New York, N.Y. : 1991)·2026
Same author

Neural responses to conflicting self- and partner-directed feedback correlate with trait emotion regulation and daily support.

Communications psychology·2026
Same author

Task-Preserving EEG Anonymization Using Latent Feature Masking.

IEEE journal of biomedical and health informatics·2026
Same author

Ghost poisoning: Making users invisible to speaker verification models.

JASA express letters·2026
Same author

Brain neuromarkers predict self- and other-related mentalizing across adult, clinical, and developmental samples.

Nature communications·2026
Same author

Temporal patterns in articulation underlying repetitions, prolongations and blocks.

Journal of fluency disorders·2026
Same journal

Diffraction perception in L-shaped rooms using virtual reality.

EURASIP journal on audio, speech, and music processing·2026
Same journal

Robust and early howling detection based on a sparsity measure.

EURASIP journal on audio, speech, and music processing·2025
Same journal

Singing to speech conversion with generative flow.

EURASIP journal on audio, speech, and music processing·2025
Same journal

Steered Response Power for Sound Source Localization: a tutorial review.

EURASIP journal on audio, speech, and music processing·2024
Same journal

A framework for the acoustic simulation of passing vehicles using variable length delay lines.

EURASIP journal on audio, speech, and music processing·2024
Same journal

Compression of room impulse responses for compact storage and fast low-latency convolution.

EURASIP journal on audio, speech, and music processing·2024
See all related articles

Related Experiment Video

Updated: Nov 17, 2025

Author Spotlight: Investigating the Impact of Emotional Prosodies on Voice Recognition and Perception
05:48

Author Spotlight: Investigating the Impact of Emotional Prosodies on Voice Recognition and Perception

Published on: August 9, 2024

1.8K

Deep multiple instance learning for foreground speech localization in ambient audio from wearable devices.

Rajat Hebbar1, Pavlos Papadopoulos1, Ramon Reyes2

  • 1Signal Analysis and Interpretation Laboratory, University of Southern California, Los Angeles, US.

EURASIP Journal on Audio, Speech, and Music Processing
|February 15, 2021
PubMed
Summary
This summary is machine-generated.

This study introduces multiple instance learning (MIL) for foreground speech detection in wearable devices, overcoming challenges with coarse audio labels and environmental noise. MIL effectively localizes speech even with limited, low-resolution annotations.

Keywords:
Foreground speech detectionMultiple instance learningWeakly labeled audioWearable audio

More Related Videos

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

4.8K
Foreign Accent and Forensic Speaker Identification in Voice Lineups: The Influence of Acoustic Features Based on Prosody
09:09

Foreign Accent and Forensic Speaker Identification in Voice Lineups: The Influence of Acoustic Features Based on Prosody

Published on: September 27, 2024

668

Related Experiment Videos

Last Updated: Nov 17, 2025

Author Spotlight: Investigating the Impact of Emotional Prosodies on Voice Recognition and Perception
05:48

Author Spotlight: Investigating the Impact of Emotional Prosodies on Voice Recognition and Perception

Published on: August 9, 2024

1.8K
Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

4.8K
Foreign Accent and Forensic Speaker Identification in Voice Lineups: The Influence of Acoustic Features Based on Prosody
09:09

Foreign Accent and Forensic Speaker Identification in Voice Lineups: The Influence of Acoustic Features Based on Prosody

Published on: September 27, 2024

668

Area of Science:

  • Audio Signal Processing
  • Machine Learning
  • Speech Recognition

Background:

  • Machine learning achieves state-of-the-art results in audio tasks, driven by large datasets and computational power.
  • A key limitation is poor generalization to real-world scenarios due to domain mismatch, particularly for foreground speech detection in wearable devices.
  • Challenges include varying environmental noise and the high cost/time associated with precise audio annotation.

Purpose of the Study:

  • To develop effective foreground speech detection models for wearable devices using coarsely labeled audio data.
  • To apply multiple instance learning (MIL) to address the challenge of low-resolution annotations in audio analysis.
  • To investigate and adapt pooling methods for densely distributed events and evaluate speech activity detection embeddings as features.

Main Methods:

  • Utilized multiple instance learning (MIL) to enable model development with lower time-resolution (coarse) annotations.
  • Applied MIL for foreground speech localization in audio streams, demonstrating both bag-level and instance-level performance.
  • Investigated various pooling methods suitable for densely occurring events and incorporated speech activity detection embeddings as input features.

Main Results:

  • Demonstrated the efficacy of MIL in localizing foreground speech using coarse audio labels.
  • Showcased successful application of MIL for foreground speech detection in challenging, noisy environments.
  • Reported performance improvements by leveraging speech activity detection embeddings as features for foreground detection.

Conclusions:

  • Multiple instance learning (MIL) provides a viable solution for foreground speech detection with coarse annotations, mitigating annotation costs.
  • The proposed methods enhance the robustness of speech detection in real-world scenarios with diverse acoustic conditions.
  • Speech activity detection embeddings offer a promising feature set for improving foreground speech detection accuracy in wearable audio devices.