Fig. 2From: A large TV dataset for speech and music activity detectionLanguage distribution in the TVSM datasetBack to article page