CoAVT integrates audio, visual, and text data for enhanced understanding.
― 7 min read
Haizhou Li
New methods improve audio-visual speaker detection in challenging environments.
― 7 min read
SEANet improves speaker isolation by reducing noise in audio processing.
― 6 min read
AdvEval exposes weaknesses in Natural Language Generation evaluation metrics.
― 6 min read
A new approach improves dialogue systems by combining topic and rhetorical structures.
― 6 min read
New model ARDiT improves text-to-speech synthesis and speech editing.
― 5 min read
A look at new methods in understanding overlapping speech during conversations.
― 8 min read
A new method improves voice conversion between languages while preserving speaker traits.
― 4 min read
A review of how data selection improves language model performance.
― 4 min read
A new framework improves connection between faces and voices, especially in noisy settings.
― 5 min read
A new method improves sound localization accuracy while ensuring data privacy.
― 4 min read
A new method for generating accented speech using text transliteration.
― 6 min read
E1 TTS transforms text into natural speech faster and more efficiently.
― 5 min read
Discover how Matryoshka embeddings improve speaker recognition efficiency and flexibility.
― 4 min read
Introducing a new model and benchmark for evaluating multi-audio tasks.
― 5 min read
New method enhances speech clarity using visual information from surroundings.
― 5 min read
Discover how emotional TTS changes communication with machines, making them more relatable.
― 6 min read