< Back to IRCAM Forum

Audiosculpt for Catalina

AudioSculpt
#9

Dear Axel,

Thank you for your reply!

I wanted to express my deepest admiration for your work: this software and SuperVP. It is in the core of my electronic composing. It is used excessively on the sound files as well as on the live processing. I could not imagine surviving without it.

Indeed, I looked at the virtual machine options, but then thought that if I want to keep my CPU reliable and not anything weird happening without me knowing (now, that’s a challenge…), I rather keep a legacy machine. I’ve now downgraded to Mojave and will keep this mid-2017 MacBook Pro as the AudioSculpt and also other older apps machine. The idea of having a virtual CPU and then performing a concert with a hugely complex Max patch, although not on the virtual machine, is raising my pulse already.

Excited to see what IrcamLAB TS2 becomes! If someone needs input on how a composer works and would like to work within this framework, I’m happy to collaborate.

I’m interested in the fully command line version of superVP. Cannot remember if we had examples on that during Cursus. If there is any tutorial, I would be happy to study that. Sorry, if this is a question someone has already answered!

Cordialement,
Maija

#10

Hi! Already asked about such a tutorial myself here: Tutorial for supervp and pm2, it looks like there is none.

I share the same sentiment with you that this forum has awesome tools to make experimental/interesting/noisy :stuck_out_tongue: music. The platform support issues are quite concerning though :frowning:

2 Likes
#11

Hi Maija, heyho666,

The best way I know in order to learn the Anasynth command-line tools is either through OpenMusic’s LISP listener, Max/MSP’s console, or AudioSculpt’s own console view. It takes some time, but it’s not that hard.

And I totally agree, Maija, virtual machines are most definitely not reliable enough for live settings… Better stick to bare metal in such cases.

All the best,
António

1 Like
#12

Dear ASusers,
I would like to express the same feelings that Maja wrote concerning the greatness and future of AS (the graphical interface), also in the name of our students, who find this software absolutely great.
It’s wonderful to know that SuperVP and pm2 will continue to work, so that one can control them from, for instance, OpenMusic, but I know of many beginners (in many countries) that cannot use AS because they have Catalina. Some computer music classes in several universities will also only subscribe to the Ircam Forum when AS will work again.
Having been in touch with the very good developers of TS2 and having tried this application, I am worried about all the compromises needed to make if fit into the commercial world, although I understand the needs for that, which would imply a substantial loss of the flexibility of the interface and a certain decrease of the quality of the sound results due to the (for the time being) exclusive usage of the real-time engine.
But I am sure that you will find a way that will allow creative composers, who will use this software in a way that professional users might never think of, to get back to this impressive and unique software.
Good luck for that!
Marco

PS: One can also access the command line in the bottom window of the AS interface, and cut&paste it into a terminal.

1 Like
#13

Hi Estudio,

As far as I know, AudioSculpt 4 is still in development, and it’s going to be quite distinct from IrcamLab TS 2 (which is indeed very limited when compared to AudioSculpt 3), and more in line with the previous generation of AudioSculpt.

As Axel from the Anasynth team hinted at above, IRCAM software development faces several (and severe) constraints, especially in terms of its workforce and development schedules. As far as I can see, it’s been like this for quite some time…

As a side note, I’m tired of all this “creatives” vs “professionals” talk we hear everywhere now… It’s just dull-brained marketspeak, it’s phony.

All the best,
António

#14

Dear all.

Many thanks for your many comments! I am really happy to have a discussion here on the Forum.
I mean that is what it is for. My reply is probably too long - sorry for this - but there were a lot of remarks.

@Maija, @heyho666, @EStudio
I am obviously very happy with your positive feedback indicating apparently that Forum software in general and AS/SuperVP/Pm2 in particular are well received by you and/or your students.

@Maija,@heyho666 - command line and tutorials: For most users command line is a show stopper - they wont even consider it. That’s why these interfaces are not teached in the IRCAM cursus and there are no tutorials. While it helps to have some understanding of the underlying signal processing algorithms, the options you find in AS dialogs are mostly one to one translations from command line parameters, so besides a more complexe presentation it is not more complexe in itself. Please don’t hesitate to pose yor questions here in the Forum. For the moment there were so few that I managed to answer all of those left handedly.

@EStudio. Many users found AudioSculpt that basically translates the command line of SuperVP into a dialog, too complexe. therefore we try to hide these … but generally we don’t drop them.

I am worried about all the compromises needed to make if fit into the commercial world, although I understand the needs for that, which would imply a substantial loss of the flexibility of the interface and a certain decrease of the quality of the sound results due to the (for the time being) exclusive usage of the real-time engine.

Concerning real time engine. This is a common misunderstanding. Together with my colleague Frederic Cornu we invested quite some time to make supervp fast, notably by means of using all the SIMD features of modern processors. I don’t deny that additionnally computers have become stronger, which certainly helps as well. So we dont rely on a real time engine in SuperVP! What you do in Max, IrcamLAB TS, on the command line, in OpenMusic (which uses command line), or in AudioSculpt, if you provide the same options you get the same result (besides rounding errors on different operation systems).It is only the interface that changes here!
There are other (non ircam) tools that favor low quality for lower cost - SuperVP (and therefore all derived products) will not do that other than letting you choose options that are faster. For example frequency domain transposition is faster but lower quality. you can select this in Max but as far as I know it is not even available in IrcamLab TS 2 !
The compromise is the other way around - you cannot run as many SuperVP objects in Max in parallel as you would like to, but we won’t compromise quality.

For friends of the command line: IrcamLAB TS2 even preserves the display of the SuperVP command line options - for the effects and the spectrogram, just click the SuperVP button, there you find the corresponding command line options. These are a litle bit different because there are
no sound files or parameter files to read, and sample rate, number of channels and so on has to be provided explicitly on the command line.

So all options are there, you find them all in the SuperVP help, but transporting a treatment from IrcamLAB TS to command line is certainly more difficult. This was a choice, on the other hand I don’t think many people ever did that.

In conclusion: there is no decrease in sound quality by design! If you have impression you have a problem with quality, then this would be a bug and you should report it!

Now flexibility: Very obviously, when it comes to signal processing and analysis and compared to AS3, there are many things missing in IrcamLAB TS2. But, I would say that the things that are already available in TS2 offer at least the same and most of the time more flexibility ! You gained “editing” (we had complaints for about 10years that this was missing), you have BPFs for all effects many of which were not available in AS3 (e.g. maxF0, sinusoidal remixing), you can tune all automation parameters (BPFs) interactively with real time feedback which was previously only possible in preview mode and only for 2 effects, there is much better multi channel support.

So I repeat, there are many parts missing, but still with respect to flexibility for me there is a considerable step forward.

Personally, I go with Maija and am curious what the developers manage to propose for the missing parts.

Best
Axel

3 Likes
Source filter synthesis in SuperVP
#15

Hi,

For anyone interested in SuperVP as a command-line tool, there’s also the SuperVP Documentation folder in AudioSculpt’s main folder. It describes every SuperVP command option, and contains lots of examples.

All the best,
António

1 Like
#16

Thanks for pointing it out! Are you referring to the “SuperVP Documentation” folder? That looks very similar to the help we would get from the command line help, I did not notice any usage examples, could you point us to the examples? I guess I am looking for beginner-level, introductory examples. It might just be me but it is not super obvious how to even specify a file path. Thanks!

#17

Hi Derya,

Yes, that’s exactly the folder I was referring to. Every SuperVP command option has at least one example on it. The file path is specified in the standard way, and always written between quotation marks (e.g., "/Users/username/foldername/filename.extension".)

If you’re not used to relying on command-line tools, you can always start with the SuperVP library for OpenMusic or the SuperVP package for Max/MSP. And like I pointed out earlier in this thread, OM’s LISP listener and Max’s console are great ways to get into the command-line tools.

All the best,
António

1 Like
#18

Thanks, I think I will start new threads for supervp command line questions but, what is wrong with the command below?

>supervp "input_file" -age -5  "output_file"

Initializing...

SuperVP::Parameter Error::leccmd::

fatal error: unexpected parameter input_file! Output filename has to be last parameter
#19

Hi Derya,

Write it like this: supervp -S"input_file" -age -5 "output_file" — and no, it’s not missing a space between -S and “input_file”, it’s really written like that.

Unless I’m missing something, that should fix it.

All the best,
António

#20

Thanks, that indeed seemed to do something, but I got another error. I am probably making an obvious mistake here - I obviously do not understand most of the technical background of the tool, I am just trying to have fun with it by tinkering :slight_smile:

Initializing...

SuperVP::Parameter Error::checkpar::

cannot use dilation factor without FFT analysis 

Done.
#21

@heyho666

Thanks, I think I will start new threads for supervp command line questions but,

I think that would be a good idea. Concerning the examples that Antonio mentioned.
Examples in the supervp documentation are not working commands. These are examples that show how the has to be used.

what is wrong with the command below?

But for starters here a few fundamentals:

In most cases you have one input sound and one output sound.

The input needs to be specified with a -S filename the output has to be the last parameter of the command line. You find the can description under “Input options” parameter -S. You see all input options by means of supervp -hi or in the html help.

For the output file please read supervp -ho the section or in the html help “Output options”-> ‘output filename’

Note that on the terminal you should avoid spaces in file names. The terminal parses arguments as space separated char sequences. If you need a space in a file you need to put ‘’ around the complete file. (This has nothing to do with supervp, it is a feature of every command line)

Further you have a few setups that work directly in the time domain. (Sample rate conversion, transposition by means of resampling, file format conversion) You can run these without any more parameters.

Note that the -age parameter is a meta parameter that is composing transposition and formant transformation to achieve the age effect in TRaX. It is not fully documented and mentioned in the doc only for completeness. It implies time stretching (see error mentions dilation) so it will not work without further parameters.

For all others you need a pair of -A (for analysis) -Z for resynthesis and in between the processing options you want to apply. Adding only -A and -Z you get the most basic phase vocoder you can imagine. Quality will be accordingly. Still the command will work.

So for many of the flags you need to run: supervp ‘your flag with option’ -S file -A -Z outfilename.
This should probably have been written in the beginning of the help.

If you want high quality you would need to refine the analysis flag (-A) to have proper formant analysis. See fft analysis in analysis parameters. All analysis are described in the section Analysis modules or on the command line with supervp -hA

Then you would need to modify window size (-M) and select a shape invariant version of the phase vocoder (-shape 1). You find these and all other options related to processing parameters under processing options (in the html doc) or as output of supervp -hp.

Best
Axel

1 Like
#22

Super helpful explanation! Will play more with these and ask in a separate thread about my questions :ok_hand:

#24

Hello Lisa,

In the details there are many differences. In a rough summary AudioSculpt does sound analysis and sound transformation, TS2 only does sound transformation. Also TS2 does not contain the sinusoidal model.

Best
Axel

#25

Dear Axel,

Just a few additions to you very exhaustive comments, very much appreciated. I did experience a different quality between supervp~ commands and the same in AS, in some extreme cases, and not very big, but perceptible. I thought it was due to the limitations of Max, but, if I come across it again, I will mention it in the Forum.
However, when doing very heavy filtering, the difference between real-time and non-real-time-and-normalized results is huge. Post-normalizing the signal takes longer, than processing it in AS (and, if the signal already is integer, results in a worse quality).
Finally, our students in Stuttgart (and several other users I know) never use more than 2 channels and do not need editing techniques, as the have sequencers that handle this kind of things much more efficiently and powerfully. So, having this available in TS2 would not be perceived as a substantial improvement (but I see that other users have other requirements). However, they regularly use the command line without problems and generate parameter files by hand and algorithmically. Loosing this control would be a great loss for many of them (me included…!).
Thank you again for this interesting exchange.
Marco

#26

Dear Marco,

My point was that there is no real time engine. Default parameters in supervp~ may have been selected differently. Notably because for transposition with the default algorithm (time domain based transposition) you have computation time linearly increasing with the transposition factor.
This is irrelevant in AS but was sesn as a show stopper in Max.

Therefore we created a transposition mode that can use frequency domain transposition to achieve computational costs that are mostly independent of the transposition, the frequency domain transposition has lower quality though. If time varying computational costs are not a problem for you, you can switch this off in the supervp~ objects. There is unfortunately a compromise to make.

So, having this available in TS2 would not be perceived as a substantial improvement

Yes - I perfectly understand. This does not help everybody.

Best
Axel

Source filter synthesis in SuperVP
#28

If you’re interested in sound analysis, OpenMusic might provide an easier way than the Max patches. Of course Max has MuBu, but OM has Pm2. The command line tools have everything and then some, but they are not user-friendly. It all depends on what you want to do with the analysis, what your usual workflow is, and what kind of interface (or lack thereof) you’re comfortable with

1 Like
#30

That’s way beyond my understanding as a mere developer, but it’s the right kind of question to ask here (maybe also try repeating that in the OpenMusic community). It sounds right to me, but I’m not sure whether you’d get what you want from OM alone, let alone how to do it. I don’t think you’re the only person thinking about that, so casting a wider net could help

#31

Dear Lisa,

In addition to Fred’s carefully well recommendations, i can add this:
pm2 and supervp in OM are non-realtime, meaning you can not use it in a performance. However, you will have full control of these wonderful binaries with their respective full capacities. The display is not as in Audiosculpt unfortunately (no sonogram), so you should rely mostly on commandline tunings as it is the case in AS. However the result is displayed in OM (in sdif, sdifedit [optional] and sound display), you can also use OM’s audio functions to display and layout markers, and such. I invite you to have a look on OMSuperVP and OMPm2 OM’s libraries here:


OM is freely downloadable here:

and is available to mostly all platforms.

The tutorial patches of these libraries are available from OM’s help menu.

Please don’t hesitate to post questions if needed in our OpenMusic disscussion group concerning OM or these libraries.

Best
K

1 Like