I did some testing myself today. I’m not really sure if what I found actually means anything yet, as there are so many different factors in what is happening.
But…
In Soundforge I generated a 1khz sinewave tone which was 44.1khz 16bit.
I then loaded the sample into Renoise 1.28 and rendered a “song” which simply replayed this sample at its base note (so the rendered song is essentially identical to the original sample), and then did the same with Renoise 1.5 RC2. I used 44.1khz 16bit Arguru’s perfect sinc when rendering.
I loaded the 3 samples (original, 1.28 render, and 1.5 render) back into Soundforge, and using the Waves PAZ Frequency analyser I analysed each one with the “[Sys] Fastest on your system” preset.
Here are the results:
Original sample:
Render from Renoise 1.28:
Render from Renoise 1.5:
The original sample and 1.28 seem to be identical. I really couldn’t notice any difference in the frequency graph. 1.5 has a VERY slight difference at the far right side of the spectrum, somewhere around 20khz. As you can see from the screenshot it only shows up as 1 pixel tall in the graph, and it is certainly not audible at all when playing back the sample. So, as I said, I am not sure this really amounts to anything, but at least from this basic analysis of the results, there appears to be a very, very slight effect.
I dunno whether this would be more or less noticeable with a different kind of sound test. It is extremely difficult to notice any changes with a more complex sound such as a drumloop or a more detailed synth sound.
Anyone have any thoughts?
Personally, I don’t find this to be a problem at all. I have not experienced any kind of high frequency distortion as you have. If I had not used this frequency analysis I would have honestly never noticed anything at all. I’m not even sure if this very slight difference will affect the overall output.
Also… when rendering at 32bit instead of 16bit from Renoise 1.5, this effect is not visible in the analysis at all. The result is as clean as the original sample. So to me it just seems like some kind of very slight artifact in the 16bit mixing routine?