Tonal vs Noise

[MeldaProduction]

Hmmm, the Eventide thing seems like just a transient splitter, same thing you can do with Transient in MXXX or even using Level crossover and transient follower modulator imho.

[jmg8]

Actually it can be used for Tonal VS Noise too. Transient processing is just one way to use it.

I really hope you are able to make this Tonal VS Noise thing happen. It is such a good idea. I can think of so many different applications and ways it can improve current FX. I would add a huge amount of power to all the Melda plugins if it was available in the crossover.

Have you had any luck with it yet?

[Soarer]

+1!

[MeldaProduction]

I didn't try yet, focused on the MTurboReverb now . We'll see in the future, have it in the to-do list for investigation

[XComposer]

Meanwhile, I am still impressed by the quality of the plug-ins of the Pitch Tech series in doing this job (see above for the link). If you have Java installed on your computer, they work perfectly in Reaper, for example. I have also Flux Trax (a plug-in) and AudioSculpt (a Mac standalone app, not a plug-in) and they are also really very good in noise - tonal separation (with more than one method: the VUV analysis of AudioSculpt is also remarkable - see http://support.ircam.fr/docs/AudioSculp ... le-EN.html), but the Pitch Tech plug-ins are impressive, you should try them.

[Alphacodex]

To have that splitting function in the Melda ecosystem would be such a great thing and could improve so many modules that are already there:
Especially I would be very interested what results we could achieve with
- just tonal pitchshifting, while keepin noise-parts unchanged
- just tonal saturation / noise unchanged
- just noise compression / tonal unchanged
- just tonal harmonies / noise unchanged
- just tonal reverberation / noise surpressed
- De-Essing just on noise part / tonal unchanged
...and that are just some of the first things I would try!!

[vectorwarrior]

Yep! I'm so excited about the possibilities of this. Everything alphacodex lists sounds awesome.

I do wonder how much control we'd need over the spectral settings to ensure decent splitting, but I'm sure Vojtech can come up with an elegant solution to this.

In other news, there's more information on TC's first structural effect here as it's just been released:
https://www.youtube.com/watch?v=nEyG3Azh2Os

[Alphacodex]

The second demo from iZotope sounds most convincing to me. The bell sound is completely removed and you get a artifact free "clock ambience" that you could layer with other sounds.

[jmg8]

The second demo from iZotope sounds most convincing to me. The bell sound is completely removed and you get a artifact free "clock ambience" that you could layer with other sounds.

Which plugin? Do you mean rx?

[Alphacodex]

Yup, RX Deconstruct ... i was referring to this video:

https://www.youtube.com/watch?v=nr6dOOi77ko

[Alphacodex]

I also tried the voiceofsnow VST plugin that was mentioned here in the thread. Although already quite cool when it comes to smoothen a reverb tail, the separation is not good enough. On vocals, a big part of the sibilances were still in the tonal part. Also when putting it on a drumloop in theoray in the tonal part only tuned toms or kicks should be heard. However there is a lot more.
So I assume it's not that easy to get a good separation. Also voiceofsnow had only one tweaking parameter (that moreover did not do a lot) to optimize the crossover-point.

Cheers,
Codex

[XComposer]

Yes, RX Deconstruct seems very remarkable! I don't have it (I really would, but it's very expensive for me…). I, too, am not satisfied with voiceofsnow (Pitch Tech and Trax are both much better, though very different from one another).
What I appreciate most in this sort of software is that the "noisy" part is not eliminated; it's only separated, but keeping it intact, in very good quality, and one uses it separately. Your RX Deconstruct example is really good in this (and also Trax and Pitch Tech). These are not to be seen only as "denoisers", there is much more to do. In fact, I often use the noisy part much more than the tonal one, in my work. Besides, a remarkable fact in the Deconstruct example is that the tonal sound is bell-like, therefore inharmonic (this means that the software works in a complex way, not just isolating the harmonic spectra from the rest).

[MeldaProduction]

The problem with Deconstruct is that it is offline, it's hard to say whether that could be performed in a plugin without a serious latency.

[Alphacodex]

I'd luckily take it with serious latency if that is the only way to get that sound quality. Maybe a quick & dirty "preview" mode with low latency and a full fledged HiQuality mode with no matter how much latency is needed (maybe even adjustable between quality and latency?)

I dont know how far you already are with MTurboReverb, but if you get that noise / tonal thing done properly you could market MTurboReverb as the hottest shit available on the market:
"Adaptive structural programmable Reverb" so ASPR Technology under the hood

[vectorwarrior]

Yeah latency will absolutely not be an issue for a feature like this as it would be used during mixing not recording. Having a preview mode would be helpful, but ultimately quality is the most important thing so I'd say do not let long latency get in the way of a spectacular feature.

For the record, rx deconstruct is probably the best tonal/noise splitter I've heard. I agree about voice of snow, it's more of a proof of concept in a cheap/free plugin... But one good thing about it is: splitting the signal into tonal and noise and then recombining again creates a very artifact free output of the original, which is probably a very desirable thing.