leonhard-behr/voxpopuli-mls-de-descriptions

Natural Language Voice Descriptions of the VoxPopuli and MLS German Datasets German read and parliamentary speech paired with its transcript, acoustic measurements, discrete German descriptor tags, and a free-text German description of the speaker's voice and recording conditions. The dataset is intended for training description-conditioned TTS models such as Parler-TTS. The data was built as part of research work. It is a random subset of the pooled German portions of VoxPopuli… See the full description on the dataset page: https://huggingface.co/datasets/leonhard-behr/voxpopuli-mls-de-descriptions.