It requires finetuning to be used for downstream tasks such as Automatic Speech Recognition (ASR), or Audio Classification.