feat(speech): deduplicated and uncensored the dataset

feat(partitions): added the remaining partitions

feat(speech): added the first partition of the dataset