EnvSDD Dataset

Please use the download scripts from https://github.com/XIAOYixuan/AUDDT/tree/yixuan-dev to download and process the dataset.

chmod +x download/get_envsdd.sh
./download/get_envsdd.sh

Label Distribution

development (178,765 samples):

• spoof: 143,012 samples
• bonafide: 35,753 samples

remain (107,259 samples):

• spoof: 107,259 samples

Data Attributes

                                                           ID                                                            path  label dataset_name
0   development/fake_audio/ATA/audioldm1/TUTSED2017Dev/b006_8   development/fake_audio/ATA/audioldm1/TUTSED2017Dev/b006_8.wav  spoof       EnvSDD
1   development/fake_audio/ATA/audioldm1/TUTSED2017Dev/b094_4   development/fake_audio/ATA/audioldm1/TUTSED2017Dev/b094_4.wav  spoof       EnvSDD
2  development/fake_audio/ATA/audioldm1/TUTSED2017Dev/b005_47  development/fake_audio/ATA/audioldm1/TUTSED2017Dev/b005_47.wav  spoof       EnvSDD

How to Import

import pandas as pd

df = pd.read_parquet("development.parquet")
print(df.head())