An attempt at creating a second "base" model. Probably broken; don't use in production.
Used voice clips from Applebloom, Applejack, Cadance, Celestia, Fluttershy, Luna, Pinkie Pie, Rainbow Dash, Rarity, Scootaloo, Starlight Glimmer, Sweetie Belle, Trixie, and Twilight Sparkle.
Total duration of audio in dataset is ~29 hours.