dB Boost with the Sample Editor

[WannitBBBad]

I hope all is well with all of you. In the past my work with the Sample Editor has been limited to sampling certain effects or sounds I couldn't get with the Stage 3. however recently I have been sampling sounds from the Stage 3 to create complete multi-instrument NSMP files for other Nord keyboards. The problem that I am running into, is that I can record individual notes of a bass synth to a wave file for example (no decay, no velocity sensitivity), normalize the level across all those notes (not normalize individual notes), and then import the wave file into the Nord Sample Editor, only to find that the output NSMP sample is not as loud as the stock Nord Samples or other Stage 3 instruments - even those I used to make the Sample. The Sample Editor will boost individual notes up to 9dB, however is it typical to have to take sounds that have already been normalized and boost them further in the Sample being created? Thanks!

[ajstan]

Do you have peaks in the samples that are preventing the normalization from boosting the main portions of the wave? Take a look at this video:

[Mr_-G-]

Hi, Audacity allows you to determine the normalisation maximum amplitude. Not sure if that is what you are after.

[WannitBBBad]

Do you have peaks in the samples that are preventing the normalization from boosting the main portions of the wave? Take a look at this video:

Hello, in this case I have a couple notes where the phasing peaks for a few notes as the note sustains, so yes there is a little room for squeezing a little more volume. To your point, I might see what can be done with compression/limiter software to bump up the overall notes without losing the punch desired. Thanks!

Hi, Audacity allows you to determine the normalisation maximum amplitude. Not sure if that is what you are after.

Thanks, I'm currently using Audacity and normalizing to 99.9% If someone loads a wave into the Sample Editor that they have already normalized, I guess the question is whether folks find that they still need to boost the level with the Sample Editor to get it up to the level of the other Nord instruments. I'm assuming that boosting a wave that's already been normalized could introduce distortion at the peaks, so I want to be cautious. Thanks again!

[Mr_-G-]

Does this happen with a particular sample set or with all samples?
I wonder if you are dealing with some peculiar waveforms. There is an old paper by Manfred Schroeder https://ieeexplore.ieee.org/document/1054411 where he showed that rearranging the phases of the components in the frequency domain can reduce amplitude peaks. Since the ear is not sensitive to fixed phase shifts that means that one can have indistinguishable sounding sounds with very different power spectra (and peak amplitudes). Maybe your samples have high amplitude transients that make it difficult to increase amplitude without saturation?
Not sure how would one go about fixing that in practice (perhaps do a fast Fourier Transform, rearrange some component phases and do an inverse Fourier Transform?). I have no idea if that that would change the nature of the sound, but I thought it was interesting to mention that such methods exist.

[WannitBBBad]

Does this happen with a particular sample set or with all samples?
I wonder if you are dealing with some peculiar waveforms. There is an old paper by Manfred Schroeder https://ieeexplore.ieee.org/document/1054411 where he showed that rearranging the phases of the components in the power spectrum can reduce amplitude peaks. Since the ear is not sensitive to fixed phase shifts that means that one can have indistinguishable sounding sounds with very different power spectra (and peak amplitudes). Maybe your samples have high amplitude transients that make it difficult to increase amplitude without saturation?
Not sure how would one go about fixing that in practice (perhaps do a fast Fourier Transform, rearrange some component phases and do an inverse Fourier Transform?). I have no idea if that that would change the nature of the sound, but I thought it was interesting to mention that such methods exist.

My EE teachers would be pleased if I used a Fourier Transform for anything After looking closer at the levels particularly with the bass synth, I have a few waveforms that are throwing the normalization off. Below is a pic of two adjacent notes and the MP3 example. As you noted, the difference is hard to hear. My overall impression going into this was that I could record notes from the Stage 3, normalize the wave file, load it into the Sample Editor, and generate an NSMP/NSMP3 file that would come out at a level comparable to the original - but that's not the case. Compression seems to be helping a bit but I'm going to dig into it further - I'll let you know what I find. Thanks for your thoughts on this!

• WH Bass Synth 1 - 2 note example.mp3

Two note example of adjacent Bass Synth notes

Synth Bass - two note example.jpg (286.94 KiB) Viewed 3361 times

[Mr_-G-]

Hi, Interesting. Do you also apply the "Remove DC offset" option and the "normalize stereo channels independently"?
I see that the 2nd note has the lower part just too close at the -1 amplitude limit while there is still some space in the top part of the wave. If you increase the amplitude it will distort (that is why it cannot be normalized further?), but if you could offset it up it might be possible to then increase the amplitude afterwards and keep everything within the amplitude space?
Let us know what you find out!

[FZiegler]

There is always this guy intervening who has no practial experience in these things (me): But how about a lowpass filter? No idea at what frequency - maybe 15, 20, 30 Hz or so?

[baekgaard]

There is always this guy intervening who has no practial experience in these things (me): But how about a lowpass filter? No idea at what frequency - maybe 15, 20, 30 Hz or so?

Close It's s high pass filter that is needed.

The waveform has some really low frequency content that throws it off to one side. It will not disappear with a DC removal, but a high pass filter maybe as low as a few Hz would help a lot! I'd put it at maybe 8-16 Hz with a 12 or even 24 dB roll off. Should fix the problem... I hope

Sent from my phone in brevity

[baekgaard]

Close It's s high pass filter that is needed.

Aehm. Thanks.
I sometimes needed that to correctly normalize a recording.