HuBERT models improve speech tasks using multiple resolutions for better performance.
― 5 min read
Computer Science - Sound
RSSLatest Articles
Latest Articles
EmoMix enables the creation of speech expressing mixed emotions with precise intensity.
― 5 min read
Discover the innovative Multi-Window Masked Autoencoder method for enhanced audio processing.
― 5 min read
A novel method merges audio and visual data to repair missing speech.
― 6 min read
Exploring methods for detecting hate speech in audio broadcasts of under-resourced languages.
― 4 min read
A new method restores lost high frequencies in historical recordings.
― 7 min read
A new model improves sound diffraction in virtual environments.
― 6 min read
Contextual biasing enhances ASR systems, improving accuracy in specialized tasks.
― 5 min read
This study presents a new system for detecting pronunciation errors in language learners.
― 6 min read
The Q A system uses self-supervised learning for innovative music rearrangement.
― 6 min read
A new method enhances text-to-speech quality and emotional expression.
― 5 min read
Techniques to reduce model size while preserving performance are emerging.
― 4 min read
New model mimics analog phasing effects with improved learning techniques.
― 5 min read
A new model reduces size while improving multilingual speech recognition.
― 6 min read
A new method improves speech recognition accuracy for African accents.
― 5 min read
A new system improves speech recognition in multi-speaker settings.
― 6 min read
LipVoicer generates clear speech from silent videos using advanced lip-reading methods.
― 5 min read
New methods aim to improve communication for individuals with dysarthria.
― 6 min read
New method improves predictions by considering multiple expert scores.
― 6 min read
A look at how Whisper handles various Arabic dialects and accents.
― 5 min read
A program combining visual and audio data to enhance video comprehension.
― 5 min read
A new method improves speech act recognition in Bengali using audio and text analysis.
― 5 min read
Research explores BERT's potential in bar-level music analysis.
― 5 min read
A new system enhances math learning at home through fun interactions.
― 6 min read
A new method enhances speech recognition models using only text data for adaptation.
― 5 min read
A new model improves melody harmonization by considering emotional factors.
― 6 min read
New methods use onomatopoeia to inspire unique dance movements.
― 5 min read
Researchers improve detection of machine-generated speech using phase information adjustments.
― 6 min read
A new approach improves speech language identification using self-supervised learning and labels.
― 6 min read
A new method enhances speech recognition for dysarthric Arabic speakers.
― 5 min read
Allophant enhances phoneme recognition for languages with limited data.
― 5 min read
Introducing SANGEET, a detailed dataset on Hindustani Classical Music.
― 4 min read
A new method aims to improve fake audio detection without losing past knowledge.
― 6 min read
A new framework enhances the study of unsupervised speech recognition systems.
― 6 min read
This project helps anyone compose music using basic beats and advanced computer methods.
― 5 min read
Self-supervised models reveal insights into phonetic and phonemic distinctions in speech.
― 5 min read
Research explores the use of speech recognition in police body camera footage analysis.
― 6 min read
A look at how computers are changing music composition.
― 4 min read
New techniques enhance emotional understanding in speech processing tasks.
― 6 min read
New model LinDiff improves speech synthesis speed and quality.
― 4 min read
A new approach to audio compression reduces file size without losing quality.
― 5 min read