I tend to think that this kind of thing is okay for personal use, whereas it wouldn't be en masse - for instance, if you were using it to power and advantage a competing music site. Others' opinions may differ.
For instance, I've previously built scrapers for banking services (American Express [https://github.com/timrogers/amex] and Lloyds TSB [https://github.com/timrogers/lloydstsb]) and the UK's university application system, UCAS (https://github.com/timrogers/ucas) and none of these organisations seem to have had any problem with it, but the library has been really gratefully received by users.