Generate audio samples first using the evaluation scripts, then place them in the public/audio directory.