IRCAM's AudioSculpt

[carrieres]

the price policy is better than before
now you can have 19 products for 240€ with the Annual Premium Access, so it's not cheap but not expensive or unfair.
audiosculpt is 180€, i don't think it's unfair, if i had a intel mac i would buy it but i will not buy the update every year but one every three year or when i have a problem fixed by an update
to compare, i have komplete 5 and updgraded to 7 and 9 ...

[Xenakios]

now you can have 19 products for 240€

Well...That is indeed significantly less expensive than it used to be. Anyway, they don't really do software for Windows, so the stuff has never interested me a lot. (I do own a Mac too these days, but that's no reason to not also want Windows versions of softwares.)

[Michael L]

Bonjour, FYI IRCAM dit:

The IRCAM Annual Individual Premium Access, will provide access to all technologies and sub-products available at Ircam Forum (up to 4 authorizations per year), and their updates for a one year period starting at the purchase date.

Once the one-year subscription date has passed, the software remains active and usable but users might experience warnings on subscription renewals.

An activation code activates all subscribed technologies at once. As a consequence, Premium users would only need to activate once to use all technologies per machine.

You can not de-activate an authorized machine. This is partially why, an “individual” account can authorize up to 4 machines. [eg you can change computers four times a year; OS updates unaffected.]

[Krakatau]

the price policy is better than before
now you can have 19 products for 240€ with the Annual Premium Access, so it's not cheap but not expensive or unfair.
audiosculpt is 180€, i don't think it's unfair, if i had a intel mac i would buy it but i will not buy the update every year but one every three year or when i have a problem fixed by an update
to compare, i have komplete 5 and updgraded to 7 and 9 ...

even less for Audiosculpt (150 Euro) ...what seems unfair to me is that you need to pay full price again, every time you change computer because the license is working for one platform only, without the possibility to move your license to another

BTW It would be the third time for me...

Too bad because this software is fantastic !

[Breeze]

Since Audiosculpt has come up again, I figured it would be a good time to ask a question I've had for some time. While I'm a dual platform user, 95% of my work is done on a PC with a much more powerful machine. My MBP is quite old and not up to the IRCAM requirements for a smooth workflow, so for now buying into its subscription model doesn't seem practical.

I've heard great things about some of IRCAM's pitch transposition and time compression/expansion algorithms. One of my great sources of frustration with all the algorithms I have access to is the way they process non-musical non-harmonic sources. Anything that has long noise sequences or long modulated noise components is pretty much limited to a max +/- 30% variation from original time & pitch before artifacts creep in and ruin the effect.

So my question is: can any of IRCAM's algorithms better this performance with such sources by a significant margin? Typically examples provided are of musical sources. A perfect test candidate for this is the sound of waves crashing on the shoreline. I sure would like to hear how the best IRCAM algorithm could deal with that at 50% or more in non-transposed time-stretching. Thanks.

[Michael L]

Demo IRCAM TS for Windows..., :shrug

[Breeze]

I have a license for TS and it performs in the same ballpark as everything else I have. It is slightly better than most, but not convincing on long stretches. Are there any time-stretching algos that are better than TS with noise-based sources in the Annual Individual Premium Access library?

[cron]

FWIW I prefer granular type stretches on noisy material. Spectral resynthesis type approaches are almost always going to change the 'grain' or 'texture' of the sound. It's not VST, but give Xenakios' Hourglass a shot for stretching noisy material. Longish overlapping grains with a slight amount of time 'jitter' can do wonders for elongating noisy sounds while preserving surface texture. It's not as precise as a straight timestretch of course, but granular can do a great job of preserving the 'character' of noisy sounds, even if the resulting stretch isn't necessarily 100% time-locked to every nuance of the original. More an elongated reinterpretation, perhaps.

Choose your trade-off

[Breeze]

Thanks for that, cron. It takes a while to get used to the interface (there's a hell of a lot of parameters there!) and I'm just scratching the surface but it has some very interesting potential. But I'd be curious to know what it would take to do time-compression/expansion with a one-knob approach for these kinds of sources... it can't be impossible!

But a big thanks to Xenakios for making this program available. Donated!

[Xenakios]

Thanks for that, cron. It takes a while to get used to the interface (there's a hell of a lot of parameters there!) and I'm just scratching the surface but it has some very interesting potential. But I'd be curious to know what it would take to do time-compression/expansion with a one-knob approach for these kinds of sources... it can't be impossible!

But a big thanks to Xenakios for making this program available. Donated!

Thanks for the donation!

Agreed, HourGlass isn't so straightforward to use for quick plain time stretching tasks. Perhaps HourGlass 2 will have a simple "couple of knobs" mode that just does time scale changing and pitch shifting, possibly also doing an analysis of the sound to adapt the processing. But HG 2 hasn't yet progressed anywhere, I am waiting if the Juce C++ framework licensing will change to a more suitable one in the coming months...

[Breeze]

Thanks for the donation!

You're welcome! Thanks for sharing your hard work, doing a great job of it (resizable GUI: YES!), and making such a flexible system available.

Agreed, HourGlass isn't so straightforward to use for quick plain time stretching tasks. Perhaps HourGlass 2 will have a simple "couple of knobs" mode that just does time scale changing and pitch shifting, possibly also doing an analysis of the sound to adapt the processing.

I remember processing sound on the Atari back in the day and some of those processes were managed offline because they were impossible to do in real time. Today, the same stuff is handled in real-time on an average system.

Maybe for this type of non-harmonic pitch tracking it might be time to start looking at complex non-realtime processes again; I think we've been spoiled by realtime processing and the thought of waiting to hear the result of complex math is unbearable. Chances are that it would require far finer adjustments and compute than are available in realtime on any current system today; but likely not tomorrow. Just a thought...

[Michael L]

Are there any time-stretching algos that are better than TS

I queried IRCAM Labs and TS is the best. Have you experimented with the Analysis settings and front panel parameters? I tweaked quite different results with an ocean wave sample.
I too am a fan of Xenakios' HourGlass; for realtime try Michael Ourednik's ArgotLunar-- nice pitch mods.

[acousmod]

Hi,

A big thumb up for Hourglass too !

There is also a very nice plugin (Windows only) with can vary the time-stretching ratio to follow the transients.
I find it a very good alternative to IRCAM's TS, with more control possibilities :
https://tonecarver.wordpress.com/tcstretch/

[cron]

(double post)

[cron]

I remember processing sound on the Atari back in the day and some of those processes were managed offline because they were impossible to do in real time. Today, the same stuff is handled in real-time on an average system.

Maybe for this type of non-harmonic pitch tracking it might be time to start looking at complex non-realtime processes again; I think we've been spoiled by realtime processing and the thought of waiting to hear the result of complex math is unbearable. Chances are that it would require far finer adjustments and compute than are available in realtime on any current system today; but likely not tomorrow. Just a thought...

The mention of Atari ST has me wondering if you were using CDP (Composers' Desktop Project). If so, that's still available for PC/Mac and has gone freeware/open source in the last year or so. Spent around £190 (IIRC) on it circa 2005 and it stands as one of the best investments I've ever made.

Definitely one to check out if you're interested in offline processing, although it's still a command line program behind the scenes of any front-end GUI (Soundloom/Soundshaper) you may use. It's a little erratic as a result, but I adore it and find its quirks well worth putting up with. Every process works much faster than real-time these days and it's an amazing way of generating interesting material quickly. Setting it loose on an entire drum samples folder and having hundreds of new drum sounds in seconds, for instance.

There are some transformations in there that just can't be achieved in real-time, and many of the processes just can't be replicated in anything else. It was my secret weapon for years. Delighted to see it set free, although disappointed (if hardly surprised) that uptake didn't spike as a result. Must get myself upgraded from version 5...

http://www.composersdesktop.com/index.html