Some time ago I've readed tons of stuff about samplerates.. so I switched from 96 to 44, but now I've made an experiment and have some questions about it.

Here it goes, the setup is:
- Nuendo 3
- TestGenerator plugin from Tools
- Waves VComp
- IXL Analyser

Pic 1: 44KHz - 1000Hz sin wave - VComp is off


Pic 2: 44KHz - 1000Hz sin wave - VComp is on, no knobs touched


Pic 3: 96KHz - 1000Hz sin wave - VComp is on, no knobs touched


well, seems pretty same here, here we go with 16000Hz sin wave:

Pic 4: 44KHz - VComp is on, no knobs touched


some additional harmonic apeared at ~12KHz..

Pic 5: 96KHz - VComp is on, no knobs touched


nothing here lower 16KHz at 96KHz.

Now lets turn VComp at Input = +13, Output +3, Lim = on, Compression ratio = 1:6

here we go:

Pic 6: 44KHz


Pic 7: 96KHz


Here we can see additional freqs plus some noise at 44KHz, and clean picture at 96KHz..

that's it. Same pictures I've got with:
- waves VComp
- waves C1
- LiquidMix compressors (some of them add much more strange harmonics at 44KHz)
- Izotope Trash compressor

at 44 and 96 works same:

- Nuendo's Dynamics compressor
- Oxford Dynamics

that's all plugins i've got.

So can somebody explain this difference?

What seems to be going on is that the compressors gain control signal contains even harmonics (which makes sense), and which are rather indifferently filtered to keep the product of the gain control signal and the audio signal within the nyquest limit.

Some filtering is obviously being done, or it would be much uglier then that, but you are seeing what look like aliasing products to me.

Incidentally, for testing compressors and the like, a two tone test is good, use say 16K and 16.001K, summed together at equal level, the result varies in level over a one second cycle that can be really good for exercising this sort of thing.

HTH.

Regards, Dan.

--------------------
Audiophiles use phono leads because they are unbalanced people!

I concur.
Compression produces harmonic distortion that, when present in the 20kHz to 40kHz region, is not adequately filtered, and so, in the 44.1kHz sampled case, is reflected back into the audible range.

It seems that two of your compressor plugins are designed to deal with this - Steinberg and, as I would expect, Oxford. The others aren't. Either due to the designers' desire to minimise processor load or to the designers' ignorence...


...or both.

Increaseing the compressor attack and especially release time should help.

Did the signal chain pass through A to D and D to A, or is it all in the digital domain? If the latter, it is likely that the problem is removed by a decent D to A converter.

--------------------
Nik
Godin, Axon, Tonelab, Repeater & the skin of my teeth!

Quote Shingles:

---

Did the signal chain pass through A to D and D to A, or is it all in the digital domain? If the latter, it is likely that the problem is removed by a decent D to A converter.

---

How could it be removed? It is in the audio passband and thus cannot be filtered out!

Regards, Dan.

--------------------
Audiophiles use phono leads because they are unbalanced people!

dmills
I made test with 16K and 16.001K, so what i should understand about compressors with it? :-)

Shingles
could you explain about that 'reflections' you mentioned?
everything is in digital domain, all plugins (incl. analyser) are inserted in one channel.

Quote alexander d:

---

dmills
I made test with 16K and 16.001K, so what i should understand about compressors with it? :-)

---

The two tone test provides a nicely amplitude modulated signal, so you can look for changes in the behaviour as the thing attacks and decays.

Quote:

---

Shingles
could you explain about that 'reflections' you mentioned?
everything is in digital domain, all plugins (incl. analyser) are inserted in one channel.

---

Aliasing can be thought of as reflecting the spectrum at the nyquest limit (Sort of).

Regards, Dan.

--------------------
Audiophiles use phono leads because they are unbalanced people!