You should be able to create this sort of sounds and still have em hit a level the DT finds agreeable. I would personally not use the sampler’s input stage for clipping distortion, but use something before it to get the sound I want, then to pad it down into a dBu range the DT does find agreeable to sample. YMMV
2 Likes
yo. it seems like it’s output of my synth indeed.
I tried to record output from A4 and no matter how hard I push the overdrive and filter resonance, DT somehow record it as best as it can.
what coudl be difference between A4 raw output and analogue semi modular synth raw out? (in this case, behringer neutron)
is there some internal limiting/compressing happening in A4 out?
I honestly disagree with this and am on czesio’s side. If monitoring thru DTII yields different results when recording than monitoring, then something is off and counter intuitive. This isn’t just about gain staging, but some choice by device designers, that excludes this particular case from being predictable.
From recordings posted one can clearly see (and hear), that there is more headroom when recording for further normalisation (I am trying to guess designer’s intentions).
Would you, Jeanne, welcome same design choice in your DAW, when printing tracks? If you overdriven some stages for desired flavour, but then - as a favour to you, the user, from designer - there would be added more bits depth for computation of sound and magically your sound would come out hi-fi. That’s my analogy and why I am with czesio on this one, despite not personally going after same results.
2 Likes
mmmmm yep, my feeling is that there’s some processing happening in DT during conversion of AD, some attenuation of signal. for me it’s learning expereince so it’s all good, but I’m just wondering why OT has much more tolerance for hot signal coming in. simply different AD converter and algorithm?
it’s a pity that I can not post comparison test now as OT is at friend’s. but I would deifnitely make some test as soon as I can.
anyway, maybe it’s also nice to have a machine which tells me that my signal is totally f**k’d and I need to calm down a bit xdxd. it’s all good it’s all good. but I’d do the comparision test 
Here is my idea in addition to Jeanne’s suggestions. Since the reason seems to be normalisation and the transient looks like being of significantly higher amplitude than the body of your sound, I would try to work on source to “normalise” the transient, making it of not greater amplitude than the rest of sound.
But I don’t think you should stop here. Since I don’t like, what DTII does to you and I am on your side as a fighter of truth - death to devices that produce different results than promised via monitoring! 
- you should evidently punish your DTII for what it does. So here is another idea. Go north! Drive the shit out of your DTII until it clips that additional headroom it has in reserves! Go 20V peak to peak, if necessary. (*) Exhaust its reserves and make it submit to your will! 
*) read in 180 BPM: “person giving this advice should not be held responsible for any damages to your device, as being evidently driven by anger or resentment and clearly not the person that should be listened to”
**) I’m not suggesting anything, but my fingers are clumsy and there might be some random characters below:
1 Like
On a serious note (I hope you can recognise my sense of humor in previous post). I don’t know exact inner workings of DTII, but I have a hypothesis based on my recent observations. I have seen one or two 32-bit interfaces, that had tremendous amounts of headroom. Unsolved mystery would be discrepancy between monitoring and recording and why is that. I can only hypothesise. Maybe, just maybe, recording and live working are done with different resources? Probably someone from Elektron would have to chime in to answer that question.
Good luck with your endeavour!
Cordially,
Norman
awesome! thanks for encouragements Norman!
I have this 1010 bluebox digital mixer somewhere and it has digital limiter/compressor at the signal chain if I remember right. so running total destruction signal through it to DT could be interesting experiment as well.
the reason why I purchased hardware sampler, as a lot of you guys would agree, is for spontaenous flow of music making, so if I have to be all technical and start to engage DAW, precisely gain stage all my toys, metering peaks…etc. all that just to sample some bits of my afternoon rage session, would not make any sense.
but if this kind of digital mixer or some kind of little processor in between devices works, then it’ll be indeed acceptable. I’ll try and post the result if anybody is interested.
thanks!!
2 Likes
You did not read what I wrote, which are centered around just technical facts.
The goal of my tips was to achieve his desired result inside the DTII, or even a better result.
But to do so, one has to know what the problem is, and what lead to the desired sound.
More Tolerance = less clipping
More Headroom = less clipping
More Transparency = less clipping
And knowing about gain staging doesn’t mean one has to use gain staging in a certain way. Personally, I use gain staging [against what people commonly recommend] for the same goals as czesio: To clip, overdrive and distort things to hell 
4 Likes
Digitakt is not good for transparency in direct sampling or resampling. Monitoring the audio is not impacted because it’s a realtime monitoring of the audio coming into the inputs, the recorder is the bit which imparts something onto that audio, that’s the bottom line. Same as processing an external synth through the mixer input, it’s not impacted unless you actually sample that audio.
I don’t have a digitakt 2 but this is the case with digitakt 1 and elektron has stated point blank that they do not intend to allow users to disable the normalization feature on DT1 or DT2. In my opinion their version of normalization does something other than just normalize the audio, otherwise the change in dynamics should not be so evident.
In my experience, if you want transparency with audio on a digitakt, you have to record your samples into another recorder, move it to your daw if it needs to be chopped up further, and then move the samples into the digitakt via the transfer software.
Anything you put into the digitakt via that method sounds great, anything you record straight into the digitakt is influenced by whatever the digitakt’s recorder and whatever the subsequent algorithmic processing is doing to it. It sounds to me like you are experiencing some extreme case of that.
If you find a way around this, I’d be interested to hear about it.
Good luck.
5 Likes
I honestly believe that I understood what you wrote, hence my recommendation to OP - quoting myself:
Here is my idea in addition to Jeanne’s suggestions
But I also equally believe that you might misunderstood or not read into core problem that OP has.
And I don’t mean that in a bad way, as miscommunication is one of most common things among usually well wishing and well meaning peers.
Back to my quote. I totally understand your suggestions and they would be my recommendations for moving further. But also - they did not solved core problem, nor did my other less serious recommendations, of not having same results while monitoring or recording thru exactly same inputs of a device. And this is what I find unacceptable and why I aligned myself with czesio. Also: this is why I used analogy of programming digital track and printing it with some imagined quirk.
I hope that this time we are better understood. But let me know, if I am too muddy and you are willing to talk further. It’s possible, thought, that it’s me who doesn’t understand something - I’m never rejecting this option.
Cheers!
I appreciate your full description, but would also like to comment regarding this part, as I think that I get, why OP is dealing with extreme case.
In most cases - that I know at least - normalisation means boosting signal to match some common ground or lowering, but still not clipped signal. And that process in most cases is rather welcome, as it gives user constant food for their music. Hell, uneven volume across patches on one synth is quite big problem for me, to be honest.
Why this case is so extreme is that OP has driven signal to oblivion, but - unexpectedly - DTII seems to have some unused in case of monitoring headroom reserved for recording. And that creates this gap between both - nomen omen - clips.
Cordially,
Norman
1 Like
There was a topic where a user had some difficulty hearing the results of the compressor on resampling, so a question was raised as to whether the internal recorder comes before the compressor in the internal signal chain. Several users tested this, and it was shown that the compressor does impact the recorded audio, however something during the normalization process it took away the compressor’s audible impact.
While this test was done with DT1, it’s certainly possible the original post is highlighting a similar behavior, but like I said I believe it’s to the extreme, and your analysis that driving the input very hot is resulting in this behavior manifesting is likely linked to the same circumstances which cause the compressor to lose impact when recorded internally.
That’s at least what came to my mind. It’s only circumstantial evidence which links these things. At it’s core, the individual user’s experience is usually the most biased factor in cases like this and so what my ears hear as the normalization impacting the audio within my own experiences, may not be audible or concerning to some others. I think that likewise, in this instance, this user was hearing something specific and tested it to the extreme to see if there were some definitive proof of the behavior.
If this is the way to prove it or not, I can’t say for sure. I do agree that it becomes more evident, the more extreme the dynamics at play are.
2 Likes
“processing algorithm” is a bit ambiguous, but yeah it’ll potentially sound “different”- a slight high freq bump, different preamps and A/D chips will have different headroom available alongside 16 bit v 24 on the OT.
The question is probably more how you can get closer to where you want to be than whether there is a difference at the extremity of use or whether it should sound crusty in exactly the same way as the OT or MD.
The whole point of gain staging is that you just understand what is happening to the audio and optimize all amplifier stages to give you what you’re looking for. If your intent is to find pockets of sound that only exist within specific gain settings in the chain, that is correct use of gain staging in that context IMO.
Gain staging doesnt always have to mean “optimizing all gain stages in the chain for minimum distortion and maximum linearity”, for me it just means you aren’t clueless about it and causing unwanted results due to ignorance of it.
Like in this particular case, “correct gain staging” would be a method which results in the intended, specifically baked distortion flavour being captured as desired… which might or might not require attenuating the composite distorted sound before hitting DT’s converters by some amount of dBu’s… but this attenuation shouldnt of course alter the timbre in any way, just clean level drop
3 Likes
You put it in way better words than I did.
1 Like
Yeah just stick something else in the resampling loop that does clip/distort.
Cassette player. Oto boum. MS20 ESP some of my favourites. In my experience not digital guitar pedals 
And gain staging real low at the start of the chain so there’s loads of hiss…
I love that gated hiss effect when using a really tired sample.
3 Likes
If it’s not possible to get the DT to clip in the exact same way as the MD/OT due to use of different DACs, is the Digitakt “incorrect”?
Preferences and design and workflows vary widely of course 
Normalization does not affect dynamics though. It’s only a change in overall volume.
Not sure if it’s possible to gain the volume back up though and have the whole thing 100% undone, or if the audio will loose a small amount of quality
Yes, I’ve heard you mention that the word has that meaning, but we’re talking about what the digitakt does to audio, regardless of what the word normalize means in a traditional context.
Were it not audible it wouldn’t come up as a point of discussion. I’m not the only person who says this and as far as I know, it is not lossy compression as you theorized but I don’t know because I’m not privy to the innerworkings of the machine, only familiar with the sounds over several years of use.
1 Like
I’d bet that nothing else is happening inside the Digitakt. A drop in volume which makes it sound different because of the loudness curve of our ears. Heck, I’d do the test myself but I don’t have a Digitakt. Is that a good reason to buy one?