Help with sample sounds

Elektron Gear Analog Rytm
#21

i’m getting similar ghost octaves using just a single tone at 6k
the freq analyser i have which is more useful is 32bit only and i’m now 64
but in Max/MSP it is clear that the wav file and the file thru AR are different in regards to ghost octaves (higher) - so i get fainter traces at 12k
interesting, needs a bit more research :+1:

edit: i think the scaling i had on amplitude was telling my eyes something my ears weren’t getting - i think the artefacts are very small in the example i tried in real terms

#22

How is your Testenvironment?
That i can do it here.
Please maybe also need help that i doit correctly.

Thank you.

#23

i’m getting similar ghost octaves using just a single tone at 6k
the freq analyser i have which is more useful is 32bit only and i’m now 64
but in Max/MSP it is clear that the wav file and the file thru AR are different in regards to ghost octaves (higher) - so i get fainter traces at 12k
interesting, needs a bit more research :+1: [/quote]
Thanks for your time on testing and input Avantronica.
I´m glad is not just my RY, but not happy at RYTM’s sample engine :frowning:
I can notice this ghost octaves as distortion on samples that has high freq material, like shakers/tops and other similar stuff. :frowning:

If I have to make a bet I´ll say that the Overdrive amount is not truly 0 even when the knob say it is.

I´ll make some more tests, and back with results soon.

#24

Test Environment and procedure:
A sampled sine tone 440Hz at 0db is used to check there is no overload at any stage inside the machines, then the 0dB sample is replaced by the reference tone (3000Khz at -10db)

Reference tone is mono for both machines, for the RYTM is at 48khz for the Octatrack is at 44.1.

I´m using a MOTU 828 MK3 as audio input, I use the same channel and same cables to record both the RYTM and the OT to ensure there is not a cable or broken A-D converter.
Takes are made on Logic Pro X or Ableton Live, then analysed in RX4/Max/Etc.

#25

Dear gbravetti,

i’m not really an expert for this.
But would like to help and learn.

Please can you upload you two Sine Wave Files.
I will sent you my E-Mail.

Thank you.

#26

For info - This time with a 2khz sin tone - 16bit48k as per AR std

Top spectrograph is file (playing concurrently) with AR output (bottom)

I had to scribble in some freq markers (linear) as there’s none(found) within Max7 WTF ?

Anyways

These particular ‘ghost’ freqs get well boosted if you mess with bit reduction, but this was off

ARfilter.png1055×388 9.91 KB

tbh - it’s extremely subtle on source material like this in an audible sense

#27

Is seems that distortion happens at Sample Level stage.

For a 440Hz sine at 0dB, distortion appears at a Sample Level param of 10 for internal mix and +16 for the direct outs.

Once you have distortion at that point, any adjustment to posterior levels won´t reduce it i.e: Amp Vol, Compressor Vol, Main Volume, trigger Velocity.

I bet that as I said before, no matter that you set overdrive to 0, the signal is getting wet with the overdrive circuit.

I don´t like this :frowning:

#28

Have you also try the seperate Outputs for the test?

#29

Have you also try the seperate Outputs for the test?
[/quote]
Yes, I just said that: "For a 440Hz sine at 0dB, distortion appears at a Sample Level param of 10 for internal mix and +16 for the direct outs."

#30

Tbh - i’m a bit lost for analytical tools at this end, or at least my older workflow is disrupted - but you are right about sample level being a factor
if i reduce the level to about 100 - those non-octave ghosts drop significantly
but - to be fair - when i look at the traces with a linear amplitude, the ‘ghosts’ are all negligible - as they are from an audio perspective on that material i tested, i.e. the graphs are more alarming to the glancing eye than ear !
i think my unfamiliarity with the tools hasn’t helped here -but that’s not to say you’re not hearing some nasties in your material or with your tests
i settled on a -3db test tone, so it had a bit of headroom to be gained up - maybe try the same to see if that keeps things more in perspective
also - maybe post a link to an A/B audio file to hear the issues at your end !

#31

Tbh - i’m a bit lost for analytical tools at this end, or at least my older workflow is disrupted - but you are right about sample level being a factor
if i reduce the level to about 100 - those non-octave ghosts drop significantly
but - to be fair - when i look at the traces with a linear amplitude, the ‘ghosts’ are all negligible - as they are from an audio perspective on that material i tested, i.e. the graphs are more alarming to the glancing eye than ear !
i think my unfamiliarity with the tools hasn’t helped here -but that’s not to say you’re not hearing some nasties in your material or with your tests
i settled on a -3db test tone, so it had a bit of headroom to be gained up - maybe try the same to see if that keeps things more in perspective
also - maybe post a link to an A/B audio file to hear the issues at your end !
[/quote]
All this started when I decided to migrate all my sampled drums from my Octatrack to my new RYTM…and they didn’t sound the same.
After checking and rechecking all my settings to see where I´m adding distortion and founding anything wrong, I decided to perform the spectrogram tests.
I will re-normalize my samples to -3db but I´m getting distortion at -10db on sine waves…
Any way, Ill check…

#32

The main problem was that samples was normalised to 0db so:

Don´t normalise your samples at 0db, use -3db or less instead.

Samples normalised to 0db will present traces of distortion and less high end.

Thanks to everybody by helping to solve this, specially to Avantronica who suggested to test with samples with less amplitude.

#33

Not sure why the OT didn´t show same amounts of distortion but I found that there was a parameter on the batch process I used to convert and normalise samples that was causing 0db samples to sound distorted.

#34

Dare I bring up again the completely analog path of the sample inside the Rytm sound engine? Also looking at the diagram of the signal path I am wondering just how do you turn off the synth engine completely. Given the electrical nature of the oscillators I am wondering also if there is maybe some residual voltage leak or interference in a very small amount that might show given the digital sample and the analog oscillator both take off on a completely analog and shared signal path.

Oh by the way what was the evil parameter that was causing this?

Also I have to command your total dedication to detail and an open mind while you sorted this out. If only half the people on internet could do the same :slight_smile: