OFFICIAL PAGE: PhonicMind
PhonicMind is an online source separation service that uses artificial intelligence, trained by deep learning. After uploading your mono or stereo audio file, for a small fee the service processes it via their artificial intelligence powered source separation. Once the process is over you get four stems: vocals, drums, bass, and the remainder. You can upload compressed or uncompressed source material and the service will provide the results in the same format that you uploaded. Examples are available at the PhonicMind website.
Screenshot courtesy of PhonicMind.com