Stereo Tool

@shckrck

I've never looked at it. So I don't know what's possible there. I guess you could use information from both volume and phase differences, but I wonder how to do that without getting too many weird artifacts... You would probably get those if you try to do this in the frequency domain. Maybe something can be done in the time domain though.

I'll need to think about this (some vague ideas are popping up now, but I'm not sure at all how they would sound).

Hans,

There's an add on for Winamp and Foobar called 'Center cut' which originally comes from a program for video processing called 'Virtualdub'- I'm sure many have heard about this one, right?

Anyway, this is still one of the best algorithms for isolating the vocals and the rears. It's also open source. But yes, this works in the frequency domain and as it is now, you have no real way of deciding how much you want to distribute to the fronts, center and rears yourself.

As for doing this stuff in the time domain (you mean fft, right?) there's a very good layout for Plogue Bidule called 'SPEC', which makes use of this. Separation here is very good and highly adjustable, only downside is that it only runs in Plogue. Another thing with FFT is that the size of the bins in the low frequencies are much larger than those higher up the band, so there's is a tendency to lose some information there which can make it sound kind of harsh sometimes. The perfect solution would be wavelets.

Anyway, it's nothing like the old stuff like you can do in Adobe and such, this is much better already.

@shckrck: Frequency domain is FFT, time domain is samples.

The thing with frequency domain is that the separation of sounds is never perfect, which means that when you manipulate the location/phase info you risk introducing 'bubbling'/MP3 ringing like artifacts. Although I think it must be possible to get rid of most of them - I still need to look into that (my current stereo image filter implementation suffers a lot from this, which is why it's been turned off by default since I made the Stereo Boost filter).

You're right, I always mix them up!

But yeah, I know what you mean, you get these weird 'chipmunk' sounds, although a lot of this can be avoided by choosing a high FFT size, like 16384 and large overlap, 16 and taking it easy on the separation a bit. But the DSP is very high at these settings so for you that will not be an option.

Do you know anything about wavelets? Have you ever tried to work with them? Theoretically this should be best of both worlds, but up till now I don't see any music applications that use them. Why is that, too difficult?

Anyway, I have a question about the Declipper. I didn't have much time the past two weeks, but today I had some time to play with it a bit more. I must say that the sound is pretty impressive and I've tried some real harsh material! Know the new album by The Black Keys? Piece of garbage, sonically speaking, but the sound really improved even with this horrible production.

However, when I experimented a bit and stacked two instances above the other (I use Plogue)) I noticed something strange; for although it was already declipped coming out of the first instance, the 2nd instance still picked up clipping after that. Why is that? I don't understand this. I mean, if it's declipped already, then why would a 2nd instance still see clipping?

That's easy to explain. If I would ONLY restore samples that are DEFINITELY clipped, I would miss a lot of clipped samples in tracks that have some form of tilt - as many do.

If I declip FAR too much, it might cause artifacts. But if I only declip a bit too much, it's usually not noticeable at all.

So, to get the best possible result, I'm also decipping some samples if I have doubts about them. And that can indeed very well happen even if they are not clipped at all - or even if they have already been declipped.

I see, makes sense. Thanks for the explanation.

You can't. And that is a fact one has to accept. There is no proper way to enforce legal usage for users who don't care about it. They would simply use a crack or a copied license file. A reputable company will respect the licensing terms and therefore wouldn't use a personal license commercially as it would be as illegal as using a pirated copy. Please don't make the mistake of thinking too much about piracy. You should better focus on the legitimate users instead. They are the ones who intend to buy a license and only they are the ones who might choke on the pricing. Therefore I support the two-license approach with a personal (about half the current price) and a commercial license (at least double the current price). Cockos, the makers of REAPER, have a comparable licensing as you can see here.

I would normally just read such posts but this one has invoked me to comment as I strongly feel this software is under valued.

Firstly, heads down to Hans for an amazing piece of software. Its very clear what you can do with this software and, cricky, where your going to go with it. I can even see such software ending up in DAB systems to 'clean up' the digital clatter of AAC+ at the receiver end.

In relation to pricing, there clearly needs to be 4 levels of licensing that are required in my opinion. These are:
- Amateur
- Professional
- Commercial
- Community/Non-Profit

The first Amateur should be at the current rate and with 8 hours time utilisation restriction. The second Professional 5 times of Amateur and with a 24 hour time utilisation restriction. The third Professional, well no less than 10 times of the base level and no time utilisation restriction. The forth Community/Non-Profit at twice the Amateur rate and no time utilisation restriction, and with strict conditions of use.

This would be a license only, and any official support contract would be additional. Of course for Commercial, you would wish to do this, and let Hans grow his business, employ people and so forth. Evolution...

A happy customer of HVZ.

After thinking a lot about this I'm now thinking about creating a separate 'hobby' license for the Winamp (!) version of the declipper. I might also make it available in the Winamp version of Stereo Tool without FM options. (Which means 2 new Winamp version license types).

Hi HVZ

The way I would use Pefect Declipper is by adding the VST plugin into Adobe Audition and processing my FLAC files. I tried this with The Strokes' "Heart in a Cage" (horrifically mastered, clipping everywhere) and it gives really great results; I am genuinely impressed.

I stream my FLACs using a Squeezebox, so if I used the Winamp plugin then I would not be able to do this. As a "regular" user, I'd just like to be able to process my music tracks, and re-save them, but the price of the licence is just too high for me.

Would it be possible to create a consumer version of the VST plugin, or is that too much to hope for?

Paul