Hey all,
First post here but I’ve been reading and enjoying the Elektronauts forums a lot, smart and helpful bunch that you are
I’ve been learning some interesting stuff lately about mixing/mastering, specifically loudness, which is all fairly new to me. I wanted to see if anyone has a particular approach to this when loading in samples? While processing samples in my DAW to put on the DT, I’ve been going mostly by ‘feel’ and I understand that the DT/Transfer app does some normalisation, but would rather know if there’s an actual guideline I should be following (and then breaking the rules! :p)
For example, I’ve seen it recommended that levels for mixing should be around -23 LUFS which leaves headroom for mastering. Is this good to aim for before importing samples, or does it matter too much as long as the sample isn’t already too hot? (i.e. close to/peaking at 0db)
Also regarding loudness/signal levels, would it be poor practice in general to compress samples (I often use the ‘glue’ compressor in Ableton) before importing, considering I may also be using the DT’s compressor on the imported samples? Would this risk clipping?
Please forgive the ignorance, but always good to learn.
Sample level defaults to 100/127 so there is a little headroom anyways. The compressor on the master track is another tool you have to create headroom if needed.
In general it seems like Elektron made it way more difficult to clip on the DT than they did with the OT and it shows. The octatrack simply loves to clip, if not in the main sample then in the fx chain unless hyper vigilant. Clipping is much less of a native feature of the DT unless you really push it.
Typically for a sampler, you wouldn’t use a broadcast reference standard like -23 LUFS to set the volume of your samples. Instead, the raw sample material is general mastered to 0db FS or just below by a smidge… (like -0.5db FS) because it is assumed that you want as much resolution in the sampler for the processing to work with… and the final stage of the sampler instrument can be brought down to where you want it in the mix.
The first big blob on the right is a 1 cycle 100 Hz sine sine wave generated at 0db FS, transfered to the DT. It is the reference so we know where 0db FS on the Digitakt shows up in my recording.
The second big blob is the Digitakt’s factory sine sample (it is tuned to C below Middle C, 130.8 Hz). Here we can see the the factory sample was also mastered to 0db FS.
However, when you master a set of samples that work together, you generally find the loudest one, set it to be 0dbFS, and then scale all the others to work around it.
For example, the first 8 hits on the left in the image are the 8 “Digit” samples in the Digitakt factory kit. You can see that that the next four are about the same, and the last three (the hats and cymbal in Digit) are all scaled down: Because they are rich in high frequencies, they sound level matched to the earlier drum samples.
So, rule of thumb is:
Master your samples to 0db FS.
If mastering a kit of sounds, master them together as a single sample so it hits 0db FS, but all the sounds feel matched in volume to you - then break them into individual parts without further adjustment of the levels.
Don’t compress the samples at all, unless the compression is to achieve the sonic timbre you’re after. Compression for mixing and mastering should always be done after the voice in the sampler (so using the Digitakt’s compressor and others outside the box).
Of course, rules are made to be broken, and your situation might dictate otherwise… but -23 LUFS seems way to quiet to me.
By the way, Transfer does not do any normalization, only sample rate conversion if needed.
Thanks, this is a super helpful response I did think that would be too quiet, and would mean having to do too much work / gain staging in the DT itself which would have its own undesirable results, so appreciate you clearing that up. I’ve done a lot of work on sample prep but didn’t need this extra headache, glad that’s gone now.
So it seems best practice is to set the loudest sample around 0db FS and then everything else relative to it.
I’ll also avoid compressing before importing to DT (unless it’s for musical reasons) - I’m just used to doing this as it feels like it brings more out of the samples when I audition them in Ableton, but I suspect DT’s compression will be enough. Just thinking of live use, I want them to be at their best and punchiest, as the aim is to perform DAWless - so I may experiment and apply some light pre-compression rather than waiting until I master the mix. Bit of compromise there I guess.
By the way I already started using ElkHerd which is an amazing bit of software tbh, easily one of the most useful things for this gear - something I dread most about the DT is needing to spend time checking back for unused samples, so I’m massively grateful for that.
I did think that would be too quiet, and would mean having to do too much work / gain staging in the DT itself which would have its own undesirable results, so appreciate you clearing that up. I’ve done a lot of work on sample prep but didn’t need this extra headache, glad that’s gone now.