If the context is academic or for software development, this is a well-known dataset for training AI to transcribe singing.
While "f500.rar" is a specific compressed file name, it most frequently points to these two high-traffic collections in the music community: 🎤 500 Free Vocal Samples by Aiva
: 500 pop songs with note-level vocal labels and corresponding YouTube links.