Hi everyone,
I just wanted to let you know that a free corpus of around 1000 hours of speech, based on LibriVox's recordings, is now available at http://www.openslr.org/12/. It's a bit like VoxForge, but about an order of magnitude larger, and (IMHO) more polished(for the gory details you can see the paper linked on the download page).
Thank you for making this possible with your contributions!