This dataset aggregates multiple German speech sources with paired transcripts.
HUI_Audio_Corpus_Clean
Bundestag ASR (Train_Dataset)
De_Distant_Speech_Data_Corpus_1415_Realtek
Stadtverwaltung Boppard Meeting Recording (Youtube)… See the full description on the dataset page:
https://huggingface.co/datasets/sarannair/german_asr_2.