Hello
I’m taking tracks and remixing them/creating stereo and stem files (MP4) for use in Traktor to DJ with.
Many are not strict with tempo so I need to correct the tempo (to save having to use a beatmap and it makes it easier to copy sections/rearrange).
My question is, could anyone give me their thoughts advice on the best workflow for speed and final audio quality.
I have two options, do the stem separation first by unmixing in Spectralayers, then bringing the tracks into Cubase and analysing or manually using the warp tool and remapping the tempo, the changing the tempo definition and making the track a consistant tempo, then editing.
Or I could do the tempo remapping on the stereo file first, then separate in Spectralayers and edit/rearrange in Cubase.
Changing the tempo is going to impact the quality of the audio I assume, so I wonder if it’s best to let Spectralayers have the original recording to work on to get the best stems?
Thanks in advance for any advice/help.
In my opinion, SpectraLayers unmixing should be done first, as it’ll have the cleanest file to work from. If there are weird artifacts from time-stretching, I presume SpectraLayers may get a little more confused than with cleaner audio.
Stretching the tempo track using the Warp tool can be done as a secondary process, as all audio tracks (stems) would still be in line with one another.
You could test the unmix after time-stretching to see what it does. Maybe do some more extreme time-stretching to see what effects it might have first, if it can still differentiate between each element (Vocals, Drums, Bass, etc.)
I find that doing stem separation first (a) results in better sounding stems and (b) improves the tempo mapping process in Cubase since it’s often easier to analyze (or work manually) with just the drum track. You also have the option to use different time stretching algorithms on different tracks, so you could, for example, use one algo for drums and a different one for vocals.
Thanks for taking the time to give me your ideas Jim & Hoffstein, I agree with you having now tried it. Especially when I’m preparing a track for DJing with that’s done to a drummer (sounds great but not easy to beatmatch for an extended period). I’m now doing the following workflow:
- Create stems in Spectralayers, full unmix track, followed by extra editing/unmixing in Spectralayers to get them clean (I’ve found this is an important stage), often ending up with more than 4 stems
- Bring into Cubase to tempo map (auto followed by manual adjusting), having separated first, it gives me either the original stereo file to analyse or the drums, then I have the stems for manual adjustment for reference.
- Save the temp definition, drop tempo map and have a fixed project tempo (I usually round up or down to a whole BPM)
- Remix/edit and export either a stereo file and stems
Now, what I haven’t figured out is a workflow for mixing. When going to stereo I can apply a limiter/mastering and then adjust in Wavelab for loudness. However, when I’m exporting the four stems, I don’t know how to apply mastering/limiting/loundness across the stems? Any thoughts?
I may end up going back to stereo files mixed as I’ll need them , as this will make it easier to move from one system to another. But, where a stem file is really useful for live variations on a mix it would be good to figure out getting the stems to have the limiter/loudness built in, again, especially for old tracks. Plus making sure that when the stems are combined they are all peaking around -1db from 0 when combined. The Native Instruments Stem Creator software I use does have limiter/compressor built in, but I’m not sure how good it is.
Where I’m not changing things around too much (I often just boost the low end/add a kick) it’s easy to apply loudenss/limiting/normalisation before I go into Spectralayers, that’s the only workaround I’ve found.
Thanks again.