There's no native way of doing this. You'll have to write some script (perhaps using pyODK or ruODK) that downloads the data and makes it available the way you'd like.
There's a feature request for naming media based on form values. Would that help solve your problem?