Latest Articles for Speech 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.

New methods are improving our ability to detect fake speech effectively.

A new method enhances ASR models for individual users using quantisation and adaptation.

New models adapt to improve speech recognition efficiency and responsiveness.

Enhancing Whisper's speech recognition for Vietnamese and other low-resource languages.

This study examines how hearing ability affects speech understanding in noisy settings.

Using k-means clustering to optimize audio data for better model training.

A method to choose the best ASR model based on audio features.

MyST aims to improve children's science learning through virtual tutoring.

A look at M2MeT 2.0 and its impact on meeting transcription.

This study examines how model compression impacts speech recognition in noisy environments.

A new model improves understanding of speech and sounds simultaneously.

Introducing new models for better speech extraction in noisy environments.

Research focuses on improving ASR systems for unsegmented audio.

Examining performance gaps in speech recognition across different genders.

LLMs enhance accuracy and error correction in speech recognition systems.

PP-MeT aims to enhance accuracy in transcribing multi-speaker meetings.

This research presents a model for improving speech clarity across different conditions.

This project aims to improve recognition of Gujarati-English mixed speech.