parczech4speech-segmented

ParCzech4Speech (Sentence-Segmented Variant) Dataset Summary ParCzech4Speech (Sentence-Segmented Variant) is a large-scale Czech speech dataset based on parliamentary recordings and official transcripts. This sentence-segmented variant is designed for speech recognition and synthesis tasks, offering clean audio-text alignment and reliable segment boundaries. It is derived from the ParCzech 4.0 corpus and AudioPSP 24.01 audio collection. Using WhisperX and Wav2Vec 2.0… See the full description on the dataset page:

Source: Hugging Face Hub (ufal/parczech4speech-segmented). Metadata imported from the dataset’s Hub tags.