metadata
license: apache-2.0
configs:
- config_name: default
data_files:
- split: train
path: data/batch_*
- split: batch_0001
path: data/batch_0001-*
- split: batch_0002
path: data/batch_0002-*
- split: batch_0003
path: data/batch_0003-*
- split: batch_0004
path: data/batch_0004-*
- split: batch_0005
path: data/batch_0005-*
- split: batch_0006
path: data/batch_0006-*
language:
- en
task_categories:
- automatic-speech-recognition
- audio-to-audio
size_categories:
- 1M<n<10M
SoundScapes (Embedded)
📌 Introduction
This dataset collects the audios and annotations from the original SoundScapes dataset (laion/soundscapes).
This dataset embeds the audio and removes it using the tar files. Only the first six batches have been employed for the making of this dataset
🙏 Acknowledgement
All credits to the Laion team and the original Soundscapes teams.