photosounder compaired to PROSONIQ ISOLATE

[vinceASPECT]

Hello Forum people,

How close is Photosounder to Prosoniq isolate as a tool?

To me, they look almost identical apart from price

Is Photosounder Multi core capable?

I am having SOME success with photosounder but wish there
was more examples a STRIPPING OUT instrument sounds from a song.

Currently i am able to strip BASSS BOOMS easily. I am studying trance
music songs.

What i want to see is examples of stripping out more TREBLE related
arrpegiator synth melody lines.

Trance music tends to be harmonically complex witha foundations
sound of organic orchestral synths. I wondered how best to remove
this foundation in a bid to get to the individual sounds in TOP END
synth melody.

I discovered that perhaps NOISE susspresion examples hold the key.

The only instrument strippping examples are HORNS and some FUNKY sond.

THese kind of look pretty standard in terms of the balance of instrument
hammonics in the mix.

DOes trance music present inherant problems for Photosounder. Is the music
too harmonically rich?

I am happy to use skype etc to communicate wiht you are go through any
tutorials.

In summary, i don't really see any concrete examples of instrument stripping
using Photosounder. For example, could a classic song be used as an example so people can relate to photosounders possiblities. Could somebody examine
say some Heaven 17 song and strip out the synth lines?

I am not able to try Pronsopniq isolate because it's MAC only and costs 349 pounds.

I realize the Melodyne with DNA is capable of a similar thing on polyphonic material of the same harmonic quality. LIke piano chords etc

it was my dream to be able to strip sounds out of existing recordings and i wanted to see some demonstration of what photosounder is capable of.

For example, photosounder is shown stripping a BASS BOOM from a song. the Kick drum....this appears to be simple and work perfectly.

How easy is it to also strip TOP END sounds like lead synth lines?

I realize that photo-sounder could be quite complex to use. There is a demonstration abotu NOISE suspression which is quite complex. I wondered if instrument stipping also often requires many steps and why are there no examples of this?

I tried the magnet tool but it appears to neeed extremely well defined barriers and thresholds with sound.

any advice would be appreciated.

I currenly work with a reasonable computer but feel Photosounder really needs a quad 2.5ghz chip.

thanks

Vince.

[A_SN]

Hello Forum people,

How close is Photosounder to Prosoniq isolate as a tool?

To me, they look almost identical apart from price

Is Photosounder Multi core capable?

I am having SOME success with photosounder but wish there
was more examples a STRIPPING OUT instrument sounds from a song.

Currently i am able to strip BASSS BOOMS easily. I am studying trance
music songs.

What i want to see is examples of stripping out more TREBLE related
arrpegiator synth melody lines.

Trance music tends to be harmonically complex witha foundations
sound of organic orchestral synths. I wondered how best to remove
this foundation in a bid to get to the individual sounds in TOP END
synth melody.

I discovered that perhaps NOISE susspresion examples hold the key.

The only instrument strippping examples are HORNS and some FUNKY sond.

THese kind of look pretty standard in terms of the balance of instrument
hammonics in the mix.

DOes trance music present inherant problems for Photosounder. Is the music
too harmonically rich?

I am happy to use skype etc to communicate wiht you are go through any
tutorials.

In summary, i don't really see any concrete examples of instrument stripping
using Photosounder. For example, could a classic song be used as an example so people can relate to photosounders possiblities. Could somebody examine
say some Heaven 17 song and strip out the synth lines?

I am not able to try Pronsopniq isolate because it's MAC only and costs 349 pounds.

I realize the Melodyne with DNA is capable of a similar thing on polyphonic material of the same harmonic quality. LIke piano chords etc

it was my dream to be able to strip sounds out of existing recordings and i wanted to see some demonstration of what photosounder is capable of.

For example, photosounder is shown stripping a BASS BOOM from a song. the Kick drum....this appears to be simple and work perfectly.

How easy is it to also strip TOP END sounds like lead synth lines?

I realize that photo-sounder could be quite complex to use. There is a demonstration abotu NOISE suspression which is quite complex. I wondered if instrument stipping also often requires many steps and why are there no examples of this?

I tried the magnet tool but it appears to neeed extremely well defined barriers and thresholds with sound.

any advice would be appreciated.

I currenly work with a reasonable computer but feel Photosounder really needs a quad 2.5ghz chip.

thanks

Vince.

I didn't really check out Sonicworx Isolate, but from watching this video http://www.youtube.com/watch?v=fazD1p1PZvg it seems it does pretty much the same as Photosounder when it comes to isolation. Actually it seems they even copied the harmonics cursor (the vertically aligned + cursors) from Photosounder . Glad to see I have some influence.

Anyway, I guess they're fairly equivalent on isolation, it seems though that Sonicworx is much slower (hence why they shy away from showing it processing anything). As you said there's also the price difference, the fact that Photosounder is both for Mac and Windows and the fact that Photosounder does a lot more than isolation (and sometimes I even like to forget it even does that as I find isolation to be a frustrating process).

Yes, Photosounder is multi-core capable. Actually, according to a Mac user with 12 cores, Photosounder and HandBrake (a video conversion program) are the only two programs capable of going over 1000% of CPU usage. Photosounder uses all the cores available when processing. No idea about Sonicworx.

Anyway, I find it funny that you'd fine bass sounds easier to isolate than treble sounds, cause it's supposed to be the other way around! Bass sounds have a terrible resolution so it's hard to see what's going on, whereas at some point treble sounds have all the resolution you need. Also your particular problem shouldn't be any different than what was demoed on the isolation videos you mentioned, as it all works the same way for harmonic content. All you have to do is figure out where in the image what you want to remove is. It helps to slow down the playback so that you might see more easily how what you see matches to what you hear. Then, there's not much more to it than depicted in the videos: select either the dark spray or the smart erase ("road roller") tool with the harmonics modifier on, possibly the magnet modifier on (if you're correctly positioned it should work correctly) and just erase the stacks of lines that represent your sound. All you have to worry about is correctly aligning each cross of the harmonics cursor with the harmonics of the sound you're trying to remove.

Also note that videos that depict instrument isolation first go through instrument removal. Isolation is an extra step from removal. Usually it's the Mask Invert operation that does the difference. Removal is also simpler because it's a more forgiving process, imperfections are harder to hear when you only subtract one instrument from a mix than when you only have one instrument left.

As for the CPU, all the videos you saw were done using a dual core CPU at 1.6 GHz, although I must admit that moving on to a quad core at 3 GHz made a lot of difference (mainly because it made the whole thing 9.5x faster. The extra cache helps a lot on top of the extra cores and the extra clock speed).

And no I don't think that the noise removal process would necessarily help you. I think that just using one of the aforementioned spray tools with the harmonics modifier on should do all you need to do. If you look at the Sonicworx Isolate video I showed above it's the equivalent process that they show.

And yes I know I need to make more tutorial videos, that's what my new marketing guy was telling me . I realise that so far I've failed to really cover the basics of Photosounder in tutorials.

[vinceASPECT]

Hello

Thanks very much for the quik response.

I realize that my comments about photosounder were not particularly accurate.

Photosounder appears to do quite a lot more than the Pronsoniq tool.

I feel that the painting of sounds is fascinating.

Also, opening existing sounds...... and editing them in photosound is really
interesting.

I will try again with the isolating instruments methods.

Yes, i imagine a quad core computer would make PS work really well.

Quad cores are pretty inexpensive now. I may get ONE just for this music
tool.

V.

[vinceASPECT]

Hello,

Been thinking of just getting this quad core computer..... and just getting on with it.

My computing habbits point to the fact that a quad core machine is handy for me.

i am running several applications at the same time.....and those applications are often ALL
multi core (multi thread)

Also, i am able to tinker under the hood and over-clock the cpu. This can result in a very
fast computer at a good price.

I was reading about muti core software tools. It seems that not ALL the tools are made equal.
For example, apparently, CUBASE tool is poorly coded for multi core. However, something
like REAPER music tool is very well coded for multi core.

I was happy to try using WORM HOLE and dedicate several computers to specific music
tasks by sending REALTIME audio over ethernet cables. However, this idea only appears
to work when the TOOLS in question are VST or VSTi tools. (i will investigate this further)

http://code.google.com/p/wormhole2/

I will check out a second hand "quad core chip" and a NEW mainboard. I Can knock up a new system
real quik here........ at low cost.

I also am interested in a music tool called CUBE by virsyn. Additive synthesis.

My interests are in Trance music production. I am not letting go of Photosounder because
EVEN just the way it can extract Bass Booms makes it worth every single penny. The types of
sounds that i hear on trance records are almost impossible to re-create. Often the producers use extremely
expensive computer software and synths. (examples are NEXUS soft synth at 800 pounds sterling)

Photosounder can give me the possibility of extracting these rare sounds for myself.

I feel that this idea alone warrants a quad core machine. My other tools will also greatly
benefit from quad cores....... and the cost of a new machine is minimal....

thanks, and i will keep looking here for advice

V.

[tahome]

How close is Photosounder to Prosoniq isolate as a tool?

I own both tools and they really don't have that much in common, although it might appear that way if you look at them from a purely visual perspective, since Isolate converts its internal representation to something more intuitive, akin to Photosounders spectrum (and doing this is an obvious choice for a tool like that).

However, Isolate has a set of highly specialized tools that can separate sounds that appear inseparable to the naked eye. You can't do this with Photoshop as graphical editing is always WYSIWYG and Photoshop (or any other graphics tool) is always limited to editing the visual representation of the sound (ie magnitude only, no access to phase and frequency etc). In sonicWORX you have tools like the magic wand - which works very differently to what the magic wand in Photoshop does - that automatically detect and remove certain sounds based on their statistical properties. Doing this with Photoshop is impossible as you're simply lacking sound specific content awareness. After all, it's an image processing tool that is [mis]used to operate on something else and you're (temporarily) losing a dimension in Photoshop since you're editing spectral magnitude only.

Photosounder is way more useful as a sound design tool as it is not as specialized as Isolate. With Isolate, removing and extracting sounds is the entire point of the application and it works incredibly well at that. I have not been able to get as good results with Photosounder, mainly because it creates a lot of artifacts when using it for that purpose and it only works as an extraction tool in rare cases (like with perfectly harmonic sounds) or when painting away large parts of the spectrum. This, and the lack of extraction quality in Melodyne DNA, was the main reason why I purchased sonicWORX in the first place btw.
Isolate is not a cheap program, but it works incredibly well and is well worth the money if you need to do high quality sound extraction on a regular basis.

On the other hand, sound design is Photosounder's forte and it is incredibly cool in this regard and I would not want to miss it for the world in my arsenal of tools! I've been a big fan of Metasynth since day one and Photosounder is a logical extension of the concept (plus I can use my beloved Photoshop for editing). But extraction is not Photosounder's main purpose and its usefulness for it is limited.

So all in all, as someone who uses both on a regular basis (and loves them both) they may look similar on a screen shot, but in reality they're quite different.

My 2 cents
--th

[A_SN]

How close is Photosounder to Prosoniq isolate as a tool?

I own both tools and they really don't have that much in common, although it might appear that way if you look at them from a purely visual perspective, since Isolate converts its internal representation to something more intuitive, akin to Photosounders spectrum (and doing this is an obvious choice for a tool like that).

However, Isolate has a set of highly specialized tools that can separate sounds that appear inseparable to the naked eye. You can't do this with Photoshop as graphical editing is always WYSIWYG and Photoshop (or any other graphics tool) is always limited to editing the visual representation of the sound (ie magnitude only, no access to phase and frequency etc). In sonicWORX you have tools like the magic wand - which works very differently to what the magic wand in Photoshop does - that automatically detect and remove certain sounds based on their statistical properties. Doing this with Photoshop is impossible as you're simply lacking sound specific content awareness. After all, it's an image processing tool that is [mis]used to operate on something else and you're (temporarily) losing a dimension in Photoshop since you're editing spectral magnitude only.

Photosounder is way more useful as a sound design tool as it is not as specialized as Isolate. With Isolate, removing and extracting sounds is the entire point of the application and it works incredibly well at that. I have not been able to get as good results with Photosounder, mainly because it creates a lot of artifacts when using it for that purpose and it only works as an extraction tool in rare cases (like with perfectly harmonic sounds) or when painting away large parts of the spectrum. This, and the lack of extraction quality in Melodyne DNA, was the main reason why I purchased sonicWORX in the first place btw.
Isolate is not a cheap program, but it works incredibly well and is well worth the money if you need to do high quality sound extraction on a regular basis.

On the other hand, sound design is Photosounder's forte and it is incredibly cool in this regard and I would not want to miss it for the world in my arsenal of tools! I've been a big fan of Metasynth since day one and Photosounder is a logical extension of the concept (plus I can use my beloved Photoshop for editing). But extraction is not Photosounder's main purpose and its usefulness for it is limited.

So all in all, as someone who uses both on a regular basis (and loves them both) they may look similar on a screen shot, but in reality they're quite different.

My 2 cents
--th

Great post, and well said. Photosounder has basic editing tools for isolation but I see isolation as much more limited and frustrating than sound design so I prefer to expand the aspects of editing useful to design than to isolation (and also I think that pure sound design is the most under-appreciated aspect of Photosounder at the moment). I hope that as I make Photosounder more usable as a sort of API to build new programs upon people will perhaps make better isolation tools using Photosounder's lossless algorithm (which is quite potent as far as isolation capabilities go). The bottom line is specialised tools will usually do better at what they're specialised than less specialised tools.

I'm curious to know what you mean by the artifacts you mentioned when using Photosounder for isolation.

[tahome]

Great post, and well said. Photosounder has basic editing tools for isolation but I see isolation as much more limited and frustrating than sound design so I prefer to expand the aspects of editing useful to design than to isolation (and also I think that pure sound design is the most under-appreciated aspect of Photosounder at the moment).

That's a real pity as this is clearly the most important aspect of it (as far as I am concerned anyway) and the main reason why I use Photosounder regularly (I do sound design for games). But it's a real problem for several products that some people just don't get it. In the studio they often ask me "can't you do extraction with Melodyne"? No you can't. It doesn't offer enough control and its audio quality for extraction is really bad. It's a great tool if you want to fix a bad note in a mix, as long as you stay within the mix. But it totally breaks down if you attempt to mute or solo single notes. Anyway, the bottom line is that every tool has its strength and weaknesses you just need to know what they are in order to use them right and get terrific results.

The bottom line is specialised tools will usually do better at what they're specialised than less specialised tools.

Agreed 100%. Isolate really has a single purpose that it excels at which is sound extraction and suppression (particularly voice and drums), but it's entirely useless if you want to get creative. Photosounder, for me, is a highly creative tool that I use to create something new out of an existing sound. I'd never use it for mastering or mix processing, of which the extraction process is part of. But I just love it for anything that requires a creative touch.

I'm curious to know what you mean by the artifacts you mentioned when using Photosounder for isolation.

Comb filter effects, warbling and phasiness mostly. They don't crop up if I do creative sound mangling with Photoshop (or, more precisely, I don't care as much and they can be part of the design process as well), but as soon as I require precise editing in order to suppress or extract individual sounds they can be a real PITA to get rid of, if at all possible.

--th

[A_SN]

I'm curious to know what you mean by the artifacts you mentioned when using Photosounder for isolation.

Comb filter effects, warbling and phasiness mostly. They don't crop up if I do creative sound mangling with Photoshop (or, more precisely, I don't care as much and they can be part of the design process as well), but as soon as I require precise editing in order to suppress or extract individual sounds they can be a real PITA to get rid of, if at all possible.

--th

Just checking, you use the lossless mode for that, right? I'm asking because you'd be surprised how many people complain about sound quality when they really just had to press that button.

[tahome]

ust checking, you use the lossless mode for that, right? I'm asking because you'd be surprised how many people complain about sound quality when they really just had to press that button.

Yes I do. I am aware that the "lossless" option keeps the phase information stored somewhere and if I want perfect reconstruction I always turn it on. Still, it doesn't help with artifacts when extracting (these artifacts are akin to a denoiser that has too high a threshold, or a low bitrate MP3).

--th

[A_SN]

ust checking, you use the lossless mode for that, right? I'm asking because you'd be surprised how many people complain about sound quality when they really just had to press that button.

Yes I do. I am aware that the "lossless" option keeps the phase information stored somewhere and if I want perfect reconstruction I always turn it on. Still, it doesn't help with artifacts when extracting (these artifacts are akin to a denoiser that has too high a threshold, or a low bitrate MP3).

--th

Ah, yeah I know what you mean. I wish I knew how other programs deal with that. That's also why Photosounder is kind of lacking at denoising. I think perhaps it because I do subtraction on linear values, so low value differences stick out like a sore thumb and make tiny blips and blops. One solution may be to do everything after having shifted the gamma up so it all looks brighter, then when it's all done shifting it down. Anyway if ever I figure out how to make it better I'll modify Photosounder accordingly.

[tahome]

Ah, yeah I know what you mean. I wish I knew how other programs deal with that. That's also why Photosounder is kind of lacking at denoising. I think perhaps it because I do subtraction on linear values, so low value differences stick out like a sore thumb and make tiny blips and blops. [...] Anyway if ever I figure out how to make it better I'll modify Photosounder accordingly.

FWIW, that was also one of the main reasons why I can't use Photosounder for extraction (aside from the specialized tools that Isolate provides). sonicWORX doesn't have these artifacts, as is apparent from their sound files: http://soundcloud.com/user7421656

I have no idea how they handle this, but according to their info they're not using a Fourier transform... could this type of artifact be specific to FFT processing?

HTH
--th

[A_SN]

Ah, yeah I know what you mean. I wish I knew how other programs deal with that. That's also why Photosounder is kind of lacking at denoising. I think perhaps it because I do subtraction on linear values, so low value differences stick out like a sore thumb and make tiny blips and blops. [...] Anyway if ever I figure out how to make it better I'll modify Photosounder accordingly.

FWIW, that was also one of the main reasons why I can't use Photosounder for extraction (aside from the specialized tools that Isolate provides). sonicWORX doesn't have these artifacts, as is apparent from their sound files: http://soundcloud.com/user7421656

I have no idea how they handle this, but according to their info they're not using a Fourier transform... could this type of artifact be specific to FFT processing?

HTH
--th

No that's definitely due to the image processing algorithm, mine is just very naive. Photosounder doesn't use FFTs as you think of them either anyway. Both Photosounder and sonicWORX use original algorithms. Although with the lossless mode you can get some type of artifact if you amplify something that shouldn't have, but I don't think that's the problem here.

Anyway, your examples illustrate why I don't bother too much with isolation. They're far from perfect, you can hear echo, bleeding from what was removed, high frequency fricatives when removing the voice, etc. I guess good enough if you just need to like change levels or remaster a bit or making it stereo, but that's a lot of work (as a developer) for a result that's pretty much bound to be imperfect. But still you should try the gamma trick I mentioned, just editing with the gamma turned up. The only problem with it is that you won't hear the result properly until you undo the gamma, but that should improve things a bit.

[tahome]

No that's definitely due to the image processing algorithm, mine is just very naive. Photosounder doesn't use FFTs as you think of them either anyway.

Are there other ways than the Cooley-Tukey (sp?) one? Just kidding, I guess there are.

Anyway, your examples illustrate why I don't bother too much with isolation. They're far from perfect, you can hear echo, bleeding from what was removed, high frequency fricatives when removing the voice, etc. I guess good enough if you just need to like change levels or remaster a bit or making it stereo, but that's a lot of work (as a developer) for a result that's pretty much bound to be imperfect.

Actually, I think they're pretty good, considering the alternatives (=none). Think about it as you judge sound quality: only five years ago you would not have thought it possible that this can even be done at all!! Melodyne DNA and sonicWORX Isolate were still Science Fiction then. I find this pretty darn amazing. It is only thanks to ingenious people like Stefan Bernsee and Pete Neubäcker that we have technology like this at our disposal today!

And as for "perfect" - there's no such thing in a digital world. There's only "good enough". Considering the difficult task I think these examples would more than qualify.

And contrary to my experience with Photosounder (and your Youtube isolation example) the Prosoniq examples exhibit no warbling artifacts whatsoever and the bit of noise that is still there won't be audible if you re-used them within a musical context (which is what you said, too, I know). Try this with Melodyne for comparison and you'll see what I mean.

If I have some time next weekend I might even put up a side-by-side sound extraction comparison of all 3 applications on a soundcloud account myself if there is enough interest.

Anyway, thanks for posting. This is an exciting application in any event and I'm sure we'll see further improvements over the months/years to come.

Cheers
--th

[A_SN]

Actually, I think they're pretty good, considering the alternatives (=none). Think about it as you judge sound quality: only five years ago you would not have thought it possible that this can even be done at all!! Melodyne DNA and sonicWORX Isolate were still Science Fiction then. I find this pretty darn amazing. It is only thanks to ingenious people like Stefan Bernsee and Pete Neubäcker that we have technology like this at our disposal today!

And as for "perfect" - there's no such thing in a digital world. There's only "good enough". Considering the difficult task I think these examples would more than qualify.

And contrary to my experience with Photosounder (and your Youtube isolation example) the Prosoniq examples exhibit no warbling artifacts whatsoever and the bit of noise that is still there won't be audible if you re-used them within a musical context (which is what you said, too, I know). Try this with Melodyne for comparison and you'll see what I mean.

If I have some time next weekend I might even put up a side-by-side sound extraction comparison of all 3 applications on a soundcloud account myself if there is enough interest.

Anyway, thanks for posting. This is an exciting application in any event and I'm sure we'll see further improvements over the months/years to come.

Cheers
--th

That's a good point, it's hard to get much of an historical perspective sometimes, it's true that it's all quite new.

Good idea about the comparison thing, although it'd be even better if you did a comparison video. And I already feel that the comparison would be unfair to Photosounder because it's the least automated of all which means the results depend a lot more on the knowledge (there are things that I myself have yet to discover), skill and dedication of the user .

What would be great would be if someone who knew ANYTHING about sound/noise isolation/removal could give me tips on how to improve it all. Like I said my algorithms are basic and unresearched, I did not research that topic, I just made basic tools that I thought would allow most things. In the meantime I'll keep working on making cooler creation tools as I've just done today http://www.kvraudio.com/forum/viewtopic ... 05#4483305

[tahome]

What would be great would be if someone who knew ANYTHING about sound/noise isolation/removal could give me tips on how to improve it all.

Well, it can't hurt to ask in the SurroundSFX.com forum. Maybe you get a reply from one of their DSP gurus...

Cheers
--th