ah, OK, the plot thickens…so you have three “similar” sounds you want to remove/ reduce?!? That is a different story. Again, select similar should be able to find the sounds if identical. That just means user has to create three different selections before asking SLP to search for similar.
See, context and full information is critical. If y’all don’t explain it in text, do you just expect the software to read your mind?
wow, yeah, if user wants SL to help, user gotta help SL a bit. Sounds like @robirdman1 needs to make the file mono (or stereo with audio in both channels) and apply some gain. I didn’t listen to the audio.
AI stems directly reflect the audio files used for training, which are mainstream, release quality music, not poor quality outliers or poor old recordings.
It took me less than a minute to properly clean the metronome ftrom your sample file using spectral editing (I didnt use Spectralayers, I used my preferred spectral editor)
Was the zip file I uploaded having low gain? I know that the previous MPEG was almost inaudible, but the original wav file that I can’t upload was not. It was originally digitized from a Teac Portastudio.
It did put more than half of the piano in Other, with the metronome.
When I got my earlier spectralyers version, I went through the manual and looked at examples and tried a lot of other things with it, frequency, selecting one area and then similar but I didn’t have much success, and it seemed too many variables and requiring much more extensive time than I could devote to get good results. So I did hope buying a later version would do more with just unmix song.
How can I tell if Cubase is using my S11 or S12 version, as I prefer using on audio sections?