The bottom line is the results speak for themselves. While the automatic separations are rarely perfect, additional spectral editing can refine the results further. Sound source separation automates the separation process somewhat by intelligently identifying elements such as vocals, drums, bass, etc. The bottom line is that spectral editing is, for the most part, a manual process that is quite time consuming and tedious. Some of the upmixes from mono to stereo which were recently released commercially made use of AudioSourceRE's SourceRE Pro software.
AudioSourceRE will be introducing new audio source separation software in a few weeks at the upcoming Audio Engineering Society (AES) Convention in New York City. The just-released RX 7 Standard and RX 7 Advanced from iZotope include the new Music Rebalance tool which makes use of sound source separation to allow you to adjust the relative levels of vocals, bass, drums, and the remainder of the music, independently even for mono source material. XTRAX STEMS will take a mono or stereo source file and return three stems: vocals, drums, and the remaining music. XTRAX STEMS from Audionamix is also an excellent example of an application that makes use of sound source separation. If you look through the research papers and presentations you'll learn that this is considered semi-supervised source separation. There are already examples of sound source separation in commercially available applications such as the ADX TRAX family of products from Audionamix which automate part of the process of separating vocals from the rest of the mono or stereo source material.
If you took my suggestion and viewed some of the research papers and presentations that I link to on the RESEARCH pages of my website you'll learn of the many approaches to sound source separation.
0 kommentar(er)
