Samples sound quieter in Reason?

[chrisizatt]

Hi,

Can anyone help me understand why, when I bring a sample into Reason (let's say for now it's an .aiff file using the NN19 Digital Sampler), if I leave all the levels at 100 - it seems quite a bit quieter than if I just play that same sample in another audio program? It just seems odd that, by default, any sample that I bring into reason isn't the same level as the actual file dictates.

I'd just like to know how to start things off with the sample as it is without any filtering.

Please help - this is doing my head in.

Thanks,
Chris.

[Aasmund1986]

Try to put on a Maximizer. Check this out: https://www.propellerheads.se/reason/effects/mclass
"maximize the volume of your Reason tracks, without crushing them or introducing other unwanted artifacts"

[chrisizatt]

Thanks for the reply. It just feels like there should be a way to get the volume of the sample to be as loud as it is in the file without having to turn anything up or add anything extra to it? It feels like a bodge rather than a fix. Makes me think that I'm losing out on quality too. Any other suggestions?

[sinnerfire]

Bit simple but have you checked the volume levels of the different applications that your using to play the samples? So i use VLC player on my mac to listen to samples on my hard drive and i usually have the volume on the player turned up so i can hear everything clear, then when i import the sample into Reasons NN-XT player and play the sample from there, it sound a bit quieter. "Answer"...Turn down the volume of the player that your using to audition the sounds on your hard drive and turn up the master volume of your computer. So the player and Reason have a more balanced sound/volume. Another thing you could try is make sure the Velocity of your midi is turned up to 127.

Hope this helps

[Dave909]

Reason lowers the levels indeed. This actually helps you as it gives you headroom to actually be able to mix sounds together and not clip the output of your interface

[WongoTheSane]

Your audio programs probably play the sound at 100%, while "level 100" in NN-XT corresponds to about 80% volume (100/128th, so 78,125% exactly). As Dave909 said, this gives you some headroom. Settings NN-XT's amp level to 127 should give you exactly the same volume as other audio programs (although it's not important in Reason, you're not losing any definition/resolution, see last week's tutorial about levels and clipping in Reason:

[Benedict]

NN-XT plays a bit differently from other devices in that it is very possible to get clipping inside the device. As Dave says,once you start layering sounds the total level increases. At the SSL that barely matters but inside the NN-XT it can get clippy and ugly really easily.

I agree tho that it is disconcerting that you have something banging, sample it off and it comes back seeming sapped. Same with adding Delays etc as Reason automatically compensates for the extra signal. Makes perfect sense but would feel nicer that as you add things, it gets beefier.

[selig]

NN-XT plays a bit differently from other devices in that it is very possible to get clipping inside the device. As Dave says,once you start layering sounds the total level increases. At the SSL that barely matters but inside the NN-XT it can get clippy and ugly really easily.

I agree tho that it is disconcerting that you have something banging, sample it off and it comes back seeming sapped. Same with adding Delays etc as Reason automatically compensates for the extra signal. Makes perfect sense but would feel nicer that as you add things, it gets beefier.

Are you SURE about that? Doesn't makes any sense that it should be doing that because all audio in Reason is 32 bit floating point. But now you've made me curious. So I decided to try to built a "torture test" and see for myself.

I loaded in a full code (just below clipping) sine wave. As you probably know, if you put release to full and master release to full, you will have inf sustain on the NNXT. I maxed out the Level to +12 dB, turned off "velocity to level", turned the master volume up to max, set the number of voices to 99. Then I played 99 notes across the keyboard, which gives a measured output of +35 dBFS. Naturally this would clip the outputs, so I lowered the gain back down under 0 dBFS (Via Selig Gain, naturally). Sounds perfect. Looks perfect (no distortion), as expected.

OK, so I have to ask, what are YOU doing to hear distortion on the NN-XT? Because everything I've tried produces none, and I feel like I've taken it MUCH further than would ever happen in normal use.

[Benedict]

Are you SURE about that? Doesn't makes any sense that it should be doing that because all audio in Reason is 32 bit floating point. But now you've made me curious. So I decided to try to built a "torture test" and see for myself.

I loaded in a full code (just below clipping) sine wave. As you probably know, if you put release to full and master release to full, you will have inf sustain on the NNXT. I maxed out the Level to +12 dB, turned off "velocity to level", turned the master volume up to max, set the number of voices to 99. Then I played 99 notes across the keyboard, which gives a measured output of +35 dBFS. Naturally this would clip the outputs, so I lowered the gain back down under 0 dBFS (Via Selig Gain, naturally). Sounds perfect. Looks perfect (no distortion), as expected.

OK, so I have to ask, what are YOU doing to hear distortion on the NN-XT? Because everything I've tried produces none, and I feel like I've taken it MUCH further than would ever happen in normal use.

Aah Mr Selig, was waiting for this. Our methods are slightly different but you may still be right.

I have several times found that using many sample layers INSIDE the NN-XT have caused clipping. Turning the gain down, on or outside of, the instrument has had no effect, but lowering all the sample's Level inside the NN-XT has solved the problem. It has commonly been with synth choirs and the like with some resonance.

I always have an M-Class Limiter across the Masters to catch overs and don't mix "in the red" so even if I am outputting the sort of levels you were in your test I shouldn't get nasty clipping sounds, esp if no meters are unhappy. When it has occurred, I have even tried Limiters after the NN-XT to sandbox it and no win. It isn't till I drop all the Sample levels that I get the clipping to go away. I have had the same in other sample players too from my Emax I & II to a VSTi called Short Circuit so maybe it is something I do.

Trying a few things right now and not able to duplicate. Any ideas welcome.

[selig]

Are you SURE about that? Doesn't makes any sense that it should be doing that because all audio in Reason is 32 bit floating point. But now you've made me curious. So I decided to try to built a "torture test" and see for myself.

I loaded in a full code (just below clipping) sine wave. As you probably know, if you put release to full and master release to full, you will have inf sustain on the NNXT. I maxed out the Level to +12 dB, turned off "velocity to level", turned the master volume up to max, set the number of voices to 99. Then I played 99 notes across the keyboard, which gives a measured output of +35 dBFS. Naturally this would clip the outputs, so I lowered the gain back down under 0 dBFS (Via Selig Gain, naturally). Sounds perfect. Looks perfect (no distortion), as expected.

OK, so I have to ask, what are YOU doing to hear distortion on the NN-XT? Because everything I've tried produces none, and I feel like I've taken it MUCH further than would ever happen in normal use.

Aah Mr Selig, was waiting for this. Our methods are slightly different but you may still be right.

I have several times found that using many sample layers INSIDE the NN-XT have caused clipping. Turning the gain down, on or outside of, the instrument has had no effect, but lowering all the sample's Level inside the NN-XT has solved the problem. It has commonly been with synth choirs and the like with some resonance.

I always have an M-Class Limiter across the Masters to catch overs and don't mix "in the red" so even if I am outputting the sort of levels you were in your test I shouldn't get nasty clipping sounds, esp if no meters are unhappy. When it has occurred, I have even tried Limiters after the NN-XT to sandbox it and no win. It isn't till I drop all the Sample levels that I get the clipping to go away. I have had the same in other sample players too from my Emax I & II to a VSTi called Short Circuit so maybe it is something I do.

Trying a few things right now and not able to duplicate. Any ideas welcome.

Hmm, resonance you say? Could be something about high resonance with the filters maybe? I'll see If I can create the issue with the filters involved. But I'm still skeptical (but willing to continue testing). So, even with full resonance I'm only able to "up" the overall level from +35 dB to +70 dBFS. And it still is not producing a lick of distortion… ;(

The thing is, it takes a doubling of voices to double the level (6 dB increase). So even if 99 voices give you + 70 dBFS, 200 voices will only give you +76 dBFS. 400 voices will give you +82 dBFS.

Now, consider there's 1500 dB dynamic range, and let's assume it's divided so that you have around 750 dB on each side of 0 dBFS. You'll run out of CPU long before you'll ever reach 750 dBFS. I've done testing where I've added 200 dB gain and then reduced it by 200 dB and it's clean, so even if there's not 750 dB above 0 dBFS, we would still have to double the number of voices over 16 more times to even reach 200 dBFS. 16 more doubles (from 400 voices) gives us over 26 MILLION voices - and we're STILL not even up to +200 dBFS.

All this to say, I'm thinking there's something else entirely going on because I'm pushing the holy crap out of the NNXT and STILL not able to hear or measure any distortion whatsoever.

Mystery still not solved - maybe we need to recreate the actual example you created where you first experienced the problem?

[Benedict]

I'd be happy to re-create. Spent some time abusing the levels myself and while I could get some scary noises I couldn't get this issue. It happens now and then and I commonly notice it in the final mixing. Next time it happens I will post the example as maybe there is something else at play.

Thanks for checking.

[WongoTheSane]

Trying a few things right now and not able to duplicate. Any ideas welcome.

There might be a specific case in which you could encounter this behaviour (lots of italics because this is pure speculation on my part, I haven't actually disassembled anything), when/if there is an effect between the NN-XT and the master that does a floating point -> 16 bits -> floating point conversion on the fly, like a bit crusher. It would depend on how the effect is programmed, but I can see several reasons why an effect would do such a conversion, bit crushing being one of them, and quick conversion from an integer 16-bits algorithm (legacy, stackoverflow/dsp copypasta...) being another. Oh, and maybe one of those REs that do maths ops like Truth (ZombieFightsShark did a tutorial about those recently), if plugged into audio instead of CV. I think it would do almost exactly what you're describing.

[Benedict]

Hmmm. Sounds interesting but not really my style of working.

[WongoTheSane]

Now, consider there's 1500 dB dynamic range, and let's assume it's divided so that you have around 750 dB on each side of 0 dBFS. You'll run out of CPU long before you'll ever reach 750 dBFS. I've done testing where I've added 200 dB gain and then reduced it by 200 dB and it's clean, so even if there's not 750 dB above 0 dBFS, we would still have to double the number of voices over 16 more times to even reach 200 dBFS. 16 more doubles (from 400 voices) gives us over 26 MILLION voices - and we're STILL not even up to +200 dBFS.

To expand on your calculations (and, for the sake of simplicity, assuming that the range is not centered but starts at 0), it would take 2^250 voices to clip, or about 1e+75 voices ("1" followed by 75 zeroes). To give some perspective, there are 1e+78 atoms in the visible universe (the thing that extends 45 billion light-years from us in every direction), and only 1e+23 in our own galaxy.

Create a Thor with 32 voices polyphony and continually press Ctrl/Cmd-D three times per second to duplicate it; when the sun dies, 5 billion years from now, you'll only have created 1e+19 voices.

Calculating the size of the loudpseaker is left as an exercise to the reader.

[GRIFTY]

Just go into the sample editor and normalize the sample, problem solved

[Marc64]

I add gain to 6.86 on the ssl mixer. it works


Skickat från min iPhone med Tapatalk

[selig]

Just go into the sample editor and normalize the sample, problem solved

Doesn't change the fact that there's 9 dB headroom built into the NNXT with the level set at 0 dB.
If you want to "normalize" the sample so it plays at 0 dBFS (assuming you've normalized it already), then adjust the NNXT sample level to +9 dB and you're good to go.

[selig]

I add gain to 6.86 on the ssl mixer. it works


Skickat från min iPhone med Tapatalk

Both NN-19 and NN-XT have 9 dB headroom, so you can't add enough gain from the SSL fader - best to use the Input Gain for this anyway IMO.

[Benedict]

Now that is good and useful info. Do you know why this was set?

[selig]

Now that is good and useful info. Do you know why this was set?

So you can play chords and not send out levels that would clip your mix with a single instrument playing. Even after that, it's best to leave additional headroom for other tracks in the mix, so I often end up turning down the master volume of many Reason instruments so they peak at around -12 dBFS max. In doing so, you'll not likely clip the mix even with 20-30 tracks or more playing.

[Benedict]

Now that is good and useful info. Do you know why this was set?

So you can play chords and not send out levels that would clip your mix with a single instrument playing. Even after that, it's best to leave additional headroom for other tracks in the mix, so I often end up turning down the master volume of many Reason instruments so they peak at around -12 dBFS max. In doing so, you'll not likely clip the mix even with 20-30 tracks or more playing.

Ok Thanks

[GRIFTY]

Just go into the sample editor and normalize the sample, problem solved

Doesn't change the fact that there's 9 dB headroom built into the NNXT with the level set at 0 dB.
If you want to "normalize" the sample so it plays at 0 dBFS (assuming you've normalized it already), then adjust the NNXT sample level to +9 dB and you're good to go.

I also didn't really know this. But I do know that when you sample sounds through the sample input reason seems to chop like five or ten dB off the sample for some annoying reason. Probably for your aforementioned chord playing answer

[nooomy]

Hi,

Can anyone help me understand why, when I bring a sample into Reason (let's say for now it's an .aiff file using the NN19 Digital Sampler), if I leave all the levels at 100 - it seems quite a bit quieter than if I just play that same sample in another audio program? It just seems odd that, by default, any sample that I bring into reason isn't the same level as the actual file dictates.

I'd just like to know how to start things off with the sample as it is without any filtering.

Please help - this is doing my head in.

Thanks,
Chris.

Have you tried normalize the sample or just simply turn up the gain? None of those effect the sample

[mreese80]

Make sure your velocity setting aren't turned down to low.

[Stranger.]

Velocity only comes into action when it's set to be used.
By default,nnxt sets a value for this,and a velocity of 1 can still play a 'full level' sample.
Redrum,kong,nn19,octorex respond in the same ways-velocity has to be set,for it to have an actual effect with midi data.
Same applies to filters etc..