sarannair/

German_ASR_New

This dataset contains only clips longer than 15 seconds from:

HUI_Audio_Corpus_Clean Bundestag ASR (Train_Dataset) De_Distant_Speech_Data_Corpus_1415_Realtek Stadtverwaltung Boppard Meeting Recording (Youtube) Stadtverwaltung Koblenz Meeting Recording (Youtube)

audio: wav audio text: transcript

Kept clips where duration > 15s (computed on the fly via streaming).

15-30s count=40840… See the full description on the dataset page: https://huggingface.co/datasets/sarannair/German_ASR_New.