Dataset · ashraq
ESC-50 environmental sound dataset
ashraq/esc50
Classic audio recording and text label dataset for environmental sound recognition.
What's inside
A popular academic corpus from the original ESC-50 repository presented at ACM Multimedia 2015. Includes audio files and corresponding text class labels, with a total volume of 1,000 to 10,000 recordings. Distributed under a non-commercial Creative Commons Attribution Non-Commercial license.
What you can do with it
- Training audio classification and environmental/natural noise recognition models
- Benchmarking audio encoders and acoustic monitoring systems
- Pretraining or validation of audio-text multimodal models
- Size
- 1K–10K rows
- Downloads this month
- 8.9K
- Likes
- 40
Models trained on it
- bioamla/ast-esc50Audio classification 75
- fatihkadim/sestemiz-unet-sisnrAudio → audio 30
- fatihkadim/sestemiz-unet-depth5Audio → audio 27
- ffuy87y/GameForge-MusicGen-SFX-v1Text → audio 23
- Akash-Sakala/audio-cnn-resnet34-esc50Audio classification 14
- BrendaTellez/SoundClassificationCNNRNNAudio classification 0
- BrendaTellez/soundsAudio classification 0
- bsye/byosm-vadVoice activity detection 0
How to load it
from datasets import load_dataset
ds = load_dataset("ashraq/esc50")