Reason Sample Devices produces different Results

[friday]

Hi All

Two weeks ago, I was at a studio from a very experienced and successful producer, mixer and mastering engineer, for an attended session. We talked a lot about technology, analog, digital… and in short, I can tell you use what gives you the best mood!

But there was a little hint that attracted me. He told me to test the Sample Players in my DAW, and he told me to be save I should simple put the samples direct in to an Audio Track (as an example he mentioned Ableton Simpler and that there in some settings the transients would not be played correctly)

So, I did the Test with different Devices in Reason and here are the Results:

First, I tested the triggering and was amazed to hear, that it is perfect! The used test sample nulled, after normalization. I have used the Redrum with the internal Sequencer against the Drum Sequencer Player, Matrix, Midi in Main Sequencer!

Then I loaded the same Sample in to the different Devices and triggered all with the Drum Sequencer Player (edit: bounced and normalized it):
• NN19 and NNXT Null's against the Reference on the Audio Track!
• Kong is very near
• Redrum and Umpf are clearly not nulling

You can download the test File under this Link: https://1drv.ms/u/s!AljUOBXZERw5upUQGGs ... Q?e=TU6ZcH

My conclusion is, I don’t know if the difference is really important in the context of music making. But in the future, I will use the NN19 or NNXT for the important Samples like Kick and Snare.

Have a great music making weekend!

[QVprod]

There’s no change in the audio file. Kong uses a mini version of NNXT (NN-Nano) to play samples but doesn’t play samples at full volume by default. The velocity is set to 80 rather than 127. Likely that similar occurs with ReDrum and Umpf.

[friday]

There’s no change in the audio file. Kong uses a mini version of NNXT (NN-Nano) to play samples but doesn’t play samples at full volume by default. The velocity is set to 80 rather than 127. Likely that similar occurs with ReDrum and Umpf.

I have normalized all bounced Samples, so the volume is not the problem! Download the file than you can experience it by yourself. There is a difference, I suspect that the sample is not played exactly from the beginning or a little late. You can see that also in the bounced waveforms, small, but when you null it against the reference in the Audio Track you hear it.

kiktestpic001.jpg (89.39 KiB) Viewed 2133 times

[diminished]

I can't find the thread but it has been observed that different devices handle samples differently. Selig wrote in that thread that he askes PH/RS about it, but I don't know what the outcome was. This was a few weeks before R11 was released.. I hope someone else is able to find it.

[friday]

I can't find the thread but it has been observed that different devices handle samples differently. Selig wrote in that thread that he askes PH/RS about it, but I don't know what the outcome was. This was a few weeks before R11 was released.. I hope someone else is able to find it.

Yes I remember that thread, but I think in the early days of the akai end emu samplers, there was also some differences. So, I can live with that. For me the important thing is, I do not need to paste the sample direct on an audio track, because NNXT and NN19 null perfect to the reference!

[diminished]

I can't find the thread but it has been observed that different devices handle samples differently. Selig wrote in that thread that he askes PH/RS about it, but I don't know what the outcome was. This was a few weeks before R11 was released.. I hope someone else is able to find it.

Yes I remember that thread, but I think in the early days of the akai end emu samplers, there was also some differences. So, I can live with that. For me the important thing is, I do not need to paste the sample direct on an audio track, because NNXT and NN19 null perfect to the reference!

That's really good to know if someone is looking for sample accuracy Thanks!

[diminished]

Found it.

Umpf - does it sound worse than redrum in default?

viewtopic.php?f=7&t=7512689

[selig]

The Umpf drums do "something" to the sample - they will cancel against each other (Club vs Retro) but not with others or audio tracks. I believe this has been reported as a bug.
Problem is, you don't know if it's 100% a sonic difference, a level difference, or a timing difference. All three will result in a failed null test, but only one of them will be audible on it's own (if it is indeed different enough to be audible). That's the problem with null tests - they don't tell you what's different, or which one of the signals is different!

I would be VERY hard pressed to pick out the difference in a blind test…

[Loque]

Did anyone ever did a null test with analog hardware?

I think ppl give a bit too much attention to something that nobody can hear.

[friday]

The Umpf drums do "something" to the sample - they will cancel against each other (Club vs Retro) but not with others or audio tracks. I believe this has been reported as a bug.
Problem is, you don't know if it's 100% a sonic difference, a level difference, or a timing difference. All three will result in a failed null test, but only one of them will be audible on it's own (if it is indeed different enough to be audible). That's the problem with null tests - they don't tell you what's different, or which one of the signals is different!

I would be VERY hard pressed to pick out the difference in a blind test…

Hi Seelig, you are right, as I wrote in one of the answers, I think Redrum cut a little from the sample start. At the first bigger wave swing, you can see the humpback comes before the reference. But I think the level can't be the difference, because it is the same single Sample from all Devices, bounced to track and then normalized, so all should have the same level. And it worked for NNXT and NN19.

For the blind test thing, I fully agree, would say only Redrum can be heard, it feels a little duller. But that can be some psych acoustic mind fu** after my test

kiktestpic002.jpg (18.6 KiB) Viewed 2077 times

[friday]

Did anyone ever did a null test with analog hardware?

I think ppl give a bit too much attention to something that nobody can hear.

Right it is not a important thing, but it doesn't hurt to know your tools exactly.

[Loque]

Did anyone ever did a null test with analog hardware?

I think ppl give a bit too much attention to something that nobody can hear.

Right it is not a important thing, but it doesn't hurt to know your tools exactly.

Agree. But choose your focus wisely.

[selig]

Hi Seelig, you are right, as I wrote in one of the answers, I think Redrum cut a little from the sample start. At the first bigger wave swing, you can see the humpback comes before the reference. But I think the level can't be the difference, because it is the same single Sample from all Devices, bounced to track and then normalized, so all should have the same level. And it worked for NNXT and NN19.

For the blind test thing, I fully agree, would say only Redrum can be heard, it feels a little duller. But that can be some psych acoustic mind fu** after my test

I got a null with ReDrum and the original audio in an audio track, so I can't see ReDrum either dulling the sound or cutting the attack.

[friday]

Hi Seelig, you are right, as I wrote in one of the answers, I think Redrum cut a little from the sample start. At the first bigger wave swing, you can see the humpback comes before the reference. But I think the level can't be the difference, because it is the same single Sample from all Devices, bounced to track and then normalized, so all should have the same level. And it worked for NNXT and NN19.

For the blind test thing, I fully agree, would say only Redrum can be heard, it feels a little duller. But that can be some psych acoustic mind fu** after my test

I got a null with ReDrum and the original audio in an audio track, so I can't see ReDrum either dulling the sound or cutting the attack.

OK, not good for my test screnario can you please share the reason file?

Edit: That didn't leave me in peace, did it again, Sample on to the Audio Track, same Sample in to the Redrum (triggered with midi + dseq + intseq) then "Bounce Mixer Channels" > "New tracks in song" > "Normalize all bounced Clips" and it did not null again? What did you do different?

[diminished]

Did anyone ever did a null test with analog hardware?

I think ppl give a bit too much attention to something that nobody can hear.

While I agree from a musical POV, that's not really the point. Sometimes you just want to look at / analyze a waveform, or send a dirac impulse through a system. If your RE/DAW messes with your dirac you're in for a bad time. Imagine if you're a developer and want to compare your VST plugin with the RE version of the same plugin. Unless you find out yourself the hard way, nobody is telling you why there is a difference on the output when the impulse sample was loaded in both, say, Umpf and NN-XT. The key word here is reproducibility.
Sample accuracy is a thing, both in development and science and it should be clear WHY and WHERE that's not given.

[Loque]

Did anyone ever did a null test with analog hardware?

I think ppl give a bit too much attention to something that nobody can hear.

While I agree from a musical POV, that's not really the point. Sometimes you just want to look at / analyze a waveform, or send a dirac impulse through a system. If your RE/DAW messes with your dirac you're in for a bad time. Imagine if you're a developer and want to compare your VST plugin with the RE version of the same plugin. Unless you find out yourself the hard way, nobody is telling you why there is a difference on the output when the impulse sample was loaded in both, say, Umpf and NN-XT. The key word here is reproducibility.
Sample accuracy is a thing, both in development and science and it should be clear WHY and WHERE that's not given.

Yea, i got it. But again, who will ever notice it? Its just a theoretically thing.

[selig]

I got a null with ReDrum and the original audio in an audio track, so I can't see ReDrum either dulling the sound or cutting the attack.

OK, not good for my test screnario can you please share the reason file?

Edit: That didn't leave me in peace, did it again, Sample on to the Audio Track, same Sample in to the Redrum (triggered with midi + dseq + intseq) then "Bounce Mixer Channels" > "New tracks in song" > "Normalize all bounced Clips" and it did not null again? What did you do different?

I used MIDI to trigger the sample, then used "bounce in place" to convert to audio.
I also just did a test with FuzzMeasure, and the biggest difference was in the impulse response, which as expected had the Umpf slightly smeared on the attack (by a little over 0.02ms). I also bounced the test file back out of Reason and it lines up perfectly with ReDrum.

[sdst]

I noticed this a year ago, I started using the nnxt it sounds better

the problem is that nnxt need more audio output.

[friday]

I used MIDI to trigger the sample, then used "bounce in place" to convert to audio.
I also just did a test with FuzzMeasure, and the biggest difference was in the impulse response, which as expected had the Umpf slightly smeared on the attack (by a little over 0.02ms). I also bounced the test file back out of Reason and it lines up perfectly with ReDrum.

Could not reproduce your experience, "bounce in place" and "bounce mixer channels > new tracks in song" plus normalizing did the same results for me?!? NN19 again nulled perfect, Redrum not. Would really love to know what I do false or if it is because of different versions? I have the newest official R11S under windows running. Here is my Reason File with the test, maybe you can check it out?

https://1drv.ms/u/s!AljUOBXZERw5upV0Kl2 ... w?e=WIoHFJ

[Spaceship]

Don’t normalize the files. Just play them back at full volume on whatever instrument you’re using. Playing a sample at low volume and then normalizing it will not phase cancel against the original sample.

[selig]

I noticed this a year ago, I started using the nnxt it sounds better

the problem is that nnxt need more audio output.

If you need more output, turn it up. The reason there's headroom is so you can play more than one note/sound at a time without automatically clipping the output.

[selig]

Don’t normalize the files. Just play them back at full volume on whatever instrument you’re using. Playing a sample at low volume and then normalizing it will not phase cancel against the original sample.

It will if the original sample was normalized…