Latest Articles for Speech Recognition

New model improves speech recognition speed and memory usage.

New methods enhance speech recognition across specific fields without extensive data.

A new model improves how computers process spoken language.

The Bayes Risk Transducer improves speech recognition efficiency and accuracy.

New dataset and framework improve spoken question answering capabilities.

Integrating metadata enhances performance in speech tasks like language identification.

This article discusses the Transducer model's real-time capabilities and recent improvements.

Research explores methods for identifying topics directly from audio recordings.

A new model connects phonetics and acoustics for better speech technology.

Research shows benefits of multiple microphones for detecting and locating speakers.

Introducing a new model for clearer speech in noisy environments.

New systems improve speaker identification using both audio and visual data.

Researchers are improving pronunciation training with new technologies for language learners.

Voice search technology evolves, addressing ASR errors for improved user experience.

A new method improves detection of fake audio using adaptive weight modification.

New model improves speech recognition in noisy environments by focusing on a single speaker.

Enhancing hybrid ASR systems for bilingual speech using grapheme units.

A new model improves speech and text alignment for better automatic recognition.

Introducing fresh metrics to assess speaker diarization accuracy in conversational AI.

New methods enhance accuracy and speed in speech recognition systems.

A new method enhances ASR performance through text data integration.

Text injection helps recognize personal information while maintaining privacy.

Radio2Text uses mmWave signals for real-time speech recognition in noisy environments.

This study enhances G2P models by focusing on error-prone areas during training.

Discover methods that improve accuracy in formant tracking for speech analysis.

New methods improve speech processing and generation in language models.

New techniques improve audio clarity in noisy environments.

New methods improve keyword spotting using available reading speech data.

A new approach enhances confidence estimation in ASR systems for better accuracy.

This study explores issues with using convnets for audio filterbank creation.

This article explores advancements in speaker diarization using language models for better accuracy.

New system enhances speech recognition using context-aware prompts.

EnCodecMAE combines self-supervised learning and audio codecs for improved audio task performance.

Introducing a flexible method for recognizing keywords in speech across languages.

PIAVE helps machines extract voices clearly, even when speakers turn their heads.

Introducing a flexible framework to enhance voice privacy research.

A new method simplifies understanding of speech classification models.

M-AUDIODEC compresses multi-channel audio while retaining speaker position and quality.

Research reveals new models to enhance voice clarity in smart earbuds.

A new method enhances robots' ability to follow spoken directions accurately.