I'd love to find a way to test this with longer audio but I don't have GPU resources and not exactly sure how to load that into the Colab. Is anyone planning on hosting or sharing a model that can be used by others to test longer form audio (for podcast transcription)?