Latest Articles for Speech Recognition

MDA enhances speech recognition by optimizing models for specific data areas.

A new method aims to enhance ASR systems for dysarthric speakers.

A new method improves computer understanding of spoken commands with fewer examples.

Enhancing speaker identification by combining sound and spoken words in audio.

A new framework improves active speaker detection using audio and visual cues.

A new method enhances general audio models for effective speech recognition.

This research addresses forgetting in AI through continual learning in spoken language understanding.

CALLS aims to improve voice assistants' ability to handle customer interactions.

New methods leverage speaker identity to improve speech recognition performance.

Using transfer learning from Czech models boosts Slovak speech recognition accuracy.

Building TTS systems for lesser-known Turkic languages using Kazakh data.

A new model improves voice isolation in noisy environments.

OpenSR enhances lip-reading models using audio data for better accuracy and accessibility.

Research reveals a model to enhance disfluency correction in speech recognition systems.

A study on how speech errors affect learning with teachable agents.

A new method improves speech recognition for names that sound alike.

New methods improve model flexibility and performance in audio tasks.

New method improves spoken language understanding without needing written transcripts.

Improving translation technology for low-resource languages like Tamasheq and Quechua.

BabySLM evaluates how well machines learn to understand speech based on children's language.

Improving systems for silent speech recognition with new techniques.

A new method for training keyword spotting models using weak supervision in noisy environments.

A new approach enhances RNN-T performance in automatic speech recognition.

Exploring methods for improved multilingual speech recognition in Indian languages.

Discover how SVVAD improves voice activity detection for better speaker verification.

A new method improves pronunciation feedback for language learners.

A new framework evaluates how well speech models adapt to specific tasks.

Research improves multilingual speech translation using semantic knowledge.

Sparq aims to improve performance in quantized neural networks with lower resource needs.

SlothSpeech reveals vulnerabilities in speech recognition systems, slowing them down significantly.

EmoMix enables the creation of speech expressing mixed emotions with precise intensity.

A new corpus for translating Cantonese audio to English text.

Discover the innovative Multi-Window Masked Autoencoder method for enhanced audio processing.

A new method enhances automatic speech recognition systems for better accuracy and adaptability.

Contextual biasing enhances ASR systems, improving accuracy in specialized tasks.

This study presents a new system for detecting pronunciation errors in language learners.

A new model reduces size while improving multilingual speech recognition.

A new system improves speech recognition in multi-speaker settings.

This study examines the benefits of merging speech processing with visual data.

A look at how Whisper handles various Arabic dialects and accents.