Pick a language. Each one holds its own sources — real human recordings, split into sentences to listen to and repeat.