
The author proposes a novel approach, BLAT, for audio-text pre-training using AudioSet tag-guided synthetic data to eliminate noise from the visual modality. The model achieves state-of-the-art performance on various downstream tasks.


coremsg

BLAT-Bootstrapping-Language-Audio-Pre-training-with-AudioSet-Tag-guided-Synthetic-Data


BLAT: Bootstrapping Language-Audio Pre-training with AudioSet Tag-guided Synthetic Data


title_rewrite


Die Studie präsentiert eine innovative Methode zur Erzeugung von hochwertigen Audio-Text-Daten durch Audio-Captioning und kontrastives Pre-Training, was zu verbesserten Leistungen in verschiedenen Audio-bezogenen Aufgaben führt.


bootstrapping-language-audio-pre-training-based-on-audioset-tag-guided-synthetic-data


Bootstrapping Language-Audio Pre-training Based on AudioSet Tag-guided Synthetic Data
