L3DAS22

L3DAS 项目旨在提供新的 3D 音频数据集和软件工具包，用于开发专为 3D 音频分析而设计的深度学习算法。为此，该项目将专注于各种沉浸式音频任务，如声音事件检测与定位、声源分离、语音识别、语音增强、音频超分辨率、声场景分类、声学回声消除和降噪等.使用 3D 录音麦克风收集的数据将通过 Python 开发的用户友好框架提供给科学界。

The L3DAS project aims to provide a novel 3D audio dataset and software toolkit for developing deep learning algorithms specifically designed for 3D audio analysis. To this end, the project will focus on a variety of immersive audio tasks, including sound event detection and localization, sound source separation, speech recognition, speech enhancement, audio super-resolution, acoustic scene classification, acoustic echo cancellation, noise reduction, etc. The data collected using 3D recording microphones will be made available to the scientific community via a user-friendly framework developed in Python.

搜集汇总

背景与挑战

背景概述

L3DAS22是一个用于3D音频分析的公开数据集，支持声音事件检测与定位、声源分离等多种沉浸式音频任务，由罗马大学等机构于2022年发布。