Idea: Wavsynth?

[stickboy]

(First, I am not an audio guy (I'm not much of a video guy either), so I admittedly have no clue what I'm talking about, and this might be a useless, stupid idea.)

Avisynth is a great and unique tool for video editing, but is there anything equivalent for audio work? Sure, Avisynth has some audio functions, but they're still video-oriented; operations are done in terms of video frames, which: is inconvenient--I would guess that timecodes are more natural for audio work may not offer the precision that some people may needWith conventional audio editors, it usually seems hard or tedious to perform operations in a reproducible, precise, and non-destructive manner.

A scripting system could solve these issues. Do any of you think a sister program to Avisynth--one targeted toward audio--would be worthwhile? Do people who work with audio care about reproducibility, precision, and non-destructiveness?

I'm not certain if it's possible to "frame-serve" WAVE files. But even if it's not, it's okay; a tool to generate a PCM WAVE file from a script would be fine too. Outputting an uncompressed file all the time might not be so bad since the storage requirements are much more practical in comparison to video.

Furthermore, I expect that a lot of programs don't use ACM (or whatever mechanisms Windows provides) to read WAVE files, choosing instead to implement their own readers, and therefore true frame-serving might not be too useful anyway.

Any thoughts?

[sh0dan]

Well - AviSynth has the full capability of being an audio processing tool. With 2.5 it has even been greatly improved, with float-point samples and (virtually) unlimited number of channels. It supports audio without video - even though it still has to go through the VFW interface. I'm not sure if there is a way of creating fake WAV-files - Anyway - SAVE As WAV is still available in VirtualDub.

However focus is mostly on the video side, as per user request, since there are no "big" demand for audio processing. So in respect to that, there has mainly been improvements in the video area.

Also the number of video plugins greatly outnumber the number of audio plugins - which in some way also reflects the usage pattern of AviSynth. I don't really believe this is just because AviSynth is video oriented - there just don't seem to be the need for it, with excellent tools like BeSweet and the like around.

I personally would like to do some more sound functions - but since I don't require them for my own work, and there isn't any big requests for it, I have stayed away from it.

[jorel]

for me,
a simple user that don't know how complex could be
develop this all seems great.

[Richard Berg]

I'm curious, do you have any suggestions on potential script syntax?

I actually do more audio work than video, and I can't imagine not having a GUI. Then again, I said the same thing about video not that long ago, so I'm willing to be convinced :)

[stickboy]

Well, you would of course need a GUI application to work alongside such a system--like how VirtualDub and Avisynth work together.

In my little world I'm imagining something like Audacity (or maybe Audacity itself) that lets you graphically apply some operations, internally generates a script, and then allows you to export that script if you so desire (similar to VirtualDub's Sylia scripting system).

I don't know how the syntax should look. I don't know a thing about audio, so I have no idea what's appropriate. I'd guess something similar to Avisynth would be adequate (with the exception that functions take times as hh:mm:ss.mmmmm instead of frame numbers, and that right end-points to ranges are always exclusive).

Of course, going in another direction and using a Python- or Matlab-like syntax might be a good idea too.

[scmccarthy]

When 2.5 came out, I mixed left-right on the left and center-surrounds on the right. The only thing that could have made that easier is if I had read it from the ac3 file directly, rather than use BeSweet to save each channel separately.

Stephen

[sh0dan]

btw, you should be able to use a framerate of 1000fps - that gives you 1ms precision.

However a GUI just for audio would be optimal - with the way AviSynth handles audio I see no reason for not doing a "complete" CoolEdit-like GUI. It's a lot of work, but it's doable.

[Malcolm]

Hi,

in September 2002 i posted a message about this topic (see here (http://forum.doom9.org/showthread.php?s=&threadid=34173))
a part of this message was: So, what about support for existing audio-filters from other applications? Cool Edit (a Shareware-Audio App) has a Filter-SDK on it's website. www.syntrillium.com

bye,
Malcolm

[sh0dan]

Some different stuff seems to be possible:

1) Import/Export of AviSynth audio in CoolEdit. I cannot tell for sure if it is possible, but it seems like it would be possible to frameserve from CoolEdit through a filter. Importing audio from an AVS-file should also be possible.

2) Using CoolEdit plugins in AviSynth. Not at all easy - several adjustments would have to made for this to be possible (as it is almost entirely GUI-based) - I'm not even sure if it is even legal. Not much of a realistic option IMO.

3) Create a filter that allowed scripts to be generated. Actually this isn't that far off. It would allow users to select samples, and then apply a filter to the selection. It probably wouldn't be with "live updates", but that should be possible at some point.

[Malcolm]

@sh0dan
could you explain your third point in detail?

i don't really care if it's cool edit filters that we can use. if there's another good / free audio-app with a decent filter system and enough existing filters then let's use this. OR: let's look at some good / important audiofilter sources and rewrite them for avisynth! (esp. denoising!!!)

i only want to avoid having multiple gigs of wav-files (orig, normalized, denoised, etc.) for a 2 hour movie on my harddrive. Basically it's the same thing as with video: You use a certain filterchain in your avs script to avoid generating multiple gigs of (huffy-) video just as an intermediate step before mpeg2/4 compression!

[WarpEnterprises]

I did a short scan on the CE API. It has a PCM interface (which is basically similar to what AviSynth offers)

and a FFT interface to work in frequency domain (this is not present in AviSynth) --> maybe this is missing here to get really further?

[E-Male]

well, as we are on this topic:

Somethign i´d like is this:

a fake wave file that just puts through the line-in
opening it in powerdvd/windvd/... would make it possible to use their audiofeatures (like upmix & dolbyheadphone) with the line in (for example to have dolbyheadphone with your normal tv program)

can someone please comment on how difficult this would be

thx
E-Male

[FredThompson]

If anyone is REALLY interested in this sort of thing, please look at http://www.csounds.com first. It's a freely-usable scripting language for audio with about 450 opcodes. Also, the editor Audacity http://audacity.sourceforge.net/ and the other links on the page I maintain at Donald Graft's site. There's a wealth of material linked there. (Yes, I know the CSound link on the page is old.)

The only noise reduction I've been able to find is the VST noise gate listed on the links page and this: http://panic.et.tudelft.nl/~costar/gramofile/

Personally, all my audio sampling is done outside the computer but this would really help a lot of folks, getting rid of that system noise squiggly crackle, I mean.

[E-Male]

csound loosk more like a synthy to me, then a wave-processor

[FredThompson]

And?...

(edit) That was a snotty reply. I shouldn't have done that.

If you poke around the csound site you'll find some GUI editors. Suppose you're editing something and want to add a drum roll or cybal crash or something like that? I don't know. Imagine anything you'd want to add music or sound effects to. What if there's a good apparent volume testing function and it could be used to bring left and right channels to the same relative volume. It's fairly common for sampled channels to have different volumes, especially with broadcast source. What if you wanted to add an echo or vibrato? Wouldn't it make sense to use a freely-available routine that's used for high quality music synthesis?

One of the major benefits of CSound is the calibre and quantity of people who use it. It gets us one step closer to a true editing suite.

Hmmm...there's an idea. I wonder what would be involved in taking one of those CSound GUIs and adding video tracks. Maybe a video clip is treated like a special type of instrument, linked by an AviSynth script. Surely there's some form of time-based composition save. Wonder if you could use it to generate a soundtrack and AviSynth script timed to it. (Not that I can do this, just wondering what might be possible since CSound is an extendable language.)

Whoa. There's an idea. I don't know the full history but AviSynth but it sure looks like something Ben did for his studies. What if some academic takes on creating an AviSynth "instrument" interface. Hmmm...

[E-Male]

just wanted to be sure
no offense intended

still looking for a way to put line-in through to programs that can play wave-files

[FredThompson]

None taken. I've had a rough couple of weeks and it sort of bubbled out right then. Sorry.

I don't understand what you are looking for. How does a WAV file redirector help with TV?

[E-Male]

the new software-dvdplayers can do an upmix to the stereo material then turn this upmix into dolbyheadphone
sounds quiet nice, i´m still working on finding the best settings
now i´d like to have this feature "live", so i can use my PC as a stereo to (surround to) dolbyheadphone converter
but this players only work with files and using the filters with graphedit doesn´t work as i like it, by far not