Sound effects are completely different to voice, which those models are trained to output.