HEAR NeurIPS 2021 Datasets (Holistic Evaluation of Audio Representations)

NOTES: * On Zenodo, please make sure you download datasets version 2021.3, not earlier versions. (2021.3 added the Vocal Imitations dataset at 48KHz. 2021.2 updated the tfds datasets.) * The datasets have different open licenses. Please see LICENSE.txt for each individual dataset's license. These are the evaluation tasks for the HEAR (Holistic Evaluation of Audio Representations) 2021 NeurIPS challenge. The aim of this challenge is to develop a general-purpose audio representation that provides a strong basis for learning in a wide variety of tasks and scenarios. The HEAR 2021 challenge invites you to create an audio embedding that is as holistic as the human ear, i.e., one that performs well across a variety of everyday domains: What approach best generalizes to a wide range of downstream audio tasks without fine-tuning? HEAR 2021 evaluates audio representations using a benchmark suite across a variety of domains, including speech, environmental sound, and music. For more information, see the HEAR 2021 website and upcoming PMLR journal article. Datasets were all normalized to a common human-readable format using hearpreprocess. Until 2022-04-01, datasets will be mirrored at data.neuralaudio.ai. This Zenodo mirror has all audio task but only at 48000Hz sampling rate. For other sampling rates (16000, 22050, 32000, 44100), please download files (requester pays) from Google Storage gs://hear2021-archive/tasks/ or AWS s3://hear2021-archive/tasks/注意事项： * 在Zenodo平台下载数据集时，请务必选择2021.3版本，请勿使用更早版本。2021.3版本新增了48KHz采样率的声音模仿（Vocal Imitations）数据集；2021.2版本更新了tfds格式的数据集。 * 本数据集集合采用各异的开放授权协议，具体请查阅各数据集对应的LICENSE.txt文件。 本数据集为NeurIPS 2021大会音频表征全方位评估（Holistic Evaluation of Audio Representations，简称HEAR）挑战赛的评估任务集。本次挑战赛的目标是研发通用型音频表征模型，为各类音频任务与场景下的学习任务提供坚实基础。 HEAR 2021挑战赛诚邀开发者构建可媲美人类听觉的全方位音频嵌入模型，即在语音、环境声与音乐等多样日常音频领域中均能实现优异性能。本次挑战赛旨在探讨的核心问题为：何种方法能够在无需微调的前提下，出色泛化至多样下游音频任务？ HEAR 2021通过覆盖多领域的基准测试套件评估音频表征，测试领域涵盖语音、环境声与音乐。如需了解更多信息，请访问HEAR 2021官方网站及即将发表于PMLR的期刊论文。 所有数据集均通过hearpreprocess工具标准化为统一的人类可读格式。 在2022年4月1日之前，数据集将在data.neuralaudio.ai平台进行镜像分发。本Zenodo镜像站点包含所有音频任务数据集，但仅提供48000Hz采样率的版本。若需获取其他采样率（16000、22050、32000、44100Hz）的数据集，请通过Google Storage（gs://hear2021-archive/tasks/）或AWS S3（s3://hear2021-archive/tasks/）下载（需由请求方承担费用）。

搜集汇总

背景与挑战

背景概述

HEAR NeurIPS 2021数据集是一个用于全面评估音频表示的综合基准数据集，旨在开发能够泛化到多种下游音频任务（如语音、环境声音和音乐）的通用音频嵌入。该数据集包含多个子任务，总计103.2 GB，所有音频已标准化为48KHz采样率，并覆盖了广泛的音频领域，适用于无需微调的音频表示学习研究。