SpeechLLMs show promise but struggle with speaker identification in conversations.
― 4 min read
Electrical Engineering and Systems Science - Audio and Speech Processing
RSSA self-supervised learning approach reduces the need for labeled audio data.
― 6 min read
Study reveals voice data's role in recognizing emotions in Spanish speakers.
― 5 min read
A new method improves speech clarity in loud environments.
― 5 min read
Innovative approaches aim to improve music quality for those with hearing loss.
― 5 min read
GenRep offers a novel approach to identifying unusual machine sounds with limited data.
― 5 min read
TF-Mamba enhances sound localization using a novel approach integrating time and frequency data.
― 5 min read
Research on modular ASR systems aims to improve performance in noisy environments.
― 4 min read
A novel method combines meaning and sound for improved emotion detection in speech.
― 6 min read
This article discusses efficient training methods for speech models using self-supervised learning.
― 4 min read
A new architecture improves sound detection across diverse environments.
― 5 min read
A new model improves music generation by focusing on individual instruments.
― 5 min read
Introducing DENSE, a method enhancing target speech extraction using dynamic embeddings.
― 6 min read
A novel method improves audio transformation while preserving melody and sound quality.
― 6 min read
This method enhances recognition accuracy for uncommon names in speech outputs.
― 6 min read
Enhancing spoken word identification through visual cues in under-resourced languages.
― 7 min read
A new model improves detection of audio deepfakes with continuous learning.
― 5 min read
An overview of audio-visual speaker diarization methods, challenges, and systems.
― 5 min read
BigCodec improves sound quality in low-bitrate audio transmission.
― 4 min read
New method improves sound capture using circular microphones for better audio quality.
― 5 min read
This article discusses the benefits of simplifying transformer models for speech tasks.
― 4 min read
Sortformer integrates speaker diarization and ASR for improved audio processing.
― 5 min read
A fresh approach to create realistic piano sounds using sound component separation.
― 8 min read
ParaEVITS improves emotional expression in TTS through natural language guidance.
― 5 min read
Learn how audio inpainting restores missing parts of signals.
― 5 min read
New methods improve understanding of spoken language through innovative dataset.
― 5 min read
New methods improve human-robot conversation by enhancing speech clarity.
― 5 min read
New methods improve access to spoken news by segmenting topics more effectively.
― 6 min read
This research analyzes Mamba's performance in speech tasks, emphasizing sound reconstruction and recognition.
― 5 min read
A new method for music tagging using few-shot learning shows promising results.
― 6 min read
FlowSep introduces a fresh method for extracting sounds using language queries.
― 5 min read
SSR-Speech offers new solutions for speech generation and editing.
― 5 min read
Advancements in AI make fake audio common, prompting the need for detection.
― 6 min read
New model enhances speech generation in diverse dialects of pitch-accent languages.
― 5 min read
A new method improves sound localization accuracy while ensuring data privacy.
― 4 min read
A new method for creating structured pop music using graph-based techniques.
― 6 min read
A new method for improving keyword spotting while retaining learned knowledge.
― 5 min read
Researchers develop a dataset to improve speech recognition and analysis techniques.
― 6 min read
SoloAudio improves sound extraction using advanced techniques and synthetic data.
― 5 min read
OpenACE provides a fair benchmark for assessing audio codecs across various conditions.
― 5 min read