Skip to main content
EURASIP Journal on Audio, Speech, and Music Processing

Table 2 Dataset description

From: Deep multiple instance learning for foreground speech localization in ambient audio from wearable devices

Dataset

Number of participants

Number of samples

% with foreground

Train/Val/Test (speakers)

AS

93

33363

21.5

75 / 9 / 9

DS

122

56091

29.7

98 / 12 / 12

Back to article page