Skip to main content

Table 7 Librispeech corpus: the various training and evaluation subsets and their size

From: Performance vs. hardware requirements in state-of-the-art automatic speech recognition

Purpose

Set

Size [h]

Training

train-clean-100

100

 

train-clean-360

360

 

train-other-500

500

Evaluation

test-clean

5.4

 

test-other

5.1