Right. At least, I've not seen it work anwyhere else...stian wrote:Cool, then they have extended the AU architecture. Still the time stretcher doesn't work anywhere else but in AU Lab, right?
--th
Dirac Timestretching anyone?
-
- KVRian
- Topic Starter
- 568 posts since 20 Nov, 2003 from Basel, Switzerland
--th
I'm the stereo chancellor
-
- KVRian
- 1325 posts since 1 Sep, 2004
Why do you feel permanently being attacked by metahome wrote:I'm not a software guy so I can't answer your question, but is it me or why are you so negative about things? Surely every problem has a solution and some clever minds will make it possible... others have done it before, too, so it can't be that hard...?
--th
Have you any problems? That wasn't an attack, nor was it asked personally to you
(Would be senseless obviously.)I asked, because that really interested me. Maybe there are already some workarrounds/experiences/concreete results there I currently merely don't see ...
By the way: That's a developer forum here.
So keep cool. Dude.
-
- KVRian
- Topic Starter
- 568 posts since 20 Nov, 2003 from Basel, Switzerland
I *am* cool, and I never said you attacked me.
I merely observed that you're constantly telling us why dirac isn't any good, and I was curious as to what made you say that. What interests *me* most is sonic quality, if there's some byte shuffling or channel instantiation or whatever techno babble is applicable required to make it work - so be it. Like you said, you're the developers, so you're supposed to tell us (the users) how cool you are in figuring out how to make it work. If you can't do it I suppose someone else will find a way. But in that case - why waste breath telling us what you *can't* do?
I don't get it...
--th
I merely observed that you're constantly telling us why dirac isn't any good, and I was curious as to what made you say that. What interests *me* most is sonic quality, if there's some byte shuffling or channel instantiation or whatever techno babble is applicable required to make it work - so be it. Like you said, you're the developers, so you're supposed to tell us (the users) how cool you are in figuring out how to make it work. If you can't do it I suppose someone else will find a way. But in that case - why waste breath telling us what you *can't* do?
I don't get it...
--th
I'm the stereo chancellor
-
- KVRian
- 1325 posts since 1 Sep, 2004
Bla bla bla... (You obviously have no clue, what you are talking about).tahome wrote:I *am* cool, and I never said you attacked me.
I merely observed that you're constantly telling us why dirac isn't any good, and I was curious as to what made you say that. What interests *me* most is sonic quality, if there's some byte shuffling or channel instantiation or whatever techno babble is applicable required to make it work - so be it. Like you said, you're the developers, so you're supposed to tell us (the users) how cool you are in figuring out how to make it work. If you can't do it I suppose someone else will find a way. But in that case - why waste breath telling us what you *can't* do?
I don't get it...
--th
I DID NEVER SAY, THAT IT ISN'T GOOD.
Last edited by useruseruser on Tue May 24, 2005 8:05 pm, edited 2 times in total.
-
- KVRian
- 1325 posts since 1 Sep, 2004
Here I prepared some samples to listen (rather than to talk permanently bullshit 
):
1. A speech sample.
I used the recommended best mode (most time domain - Lambda1) for that kind with highest possible quality. It was (compared to the frequency domain calculation) relatively fast, but much too slow for realtime, I think.
speech 50% time stretched
speech original
speech 200% time stretched
2. A music sample.
I used the recommended best mode (most frequency domain - Lambda5) with highest possible quality (see next comment).
Notes:
- I got crashes, when using frequency domain (Lambda6) which were caused inside the
binary library (so out of my focus).
- Frequency domain processing needed extraordinary long time (over 5 minutes!!!) to process on those small files.
music 50% time stretched
music original
music 200% time stretched
No. I finally don't see the hype.
(But I do NOT say, that it is bad.)
):1. A speech sample.
I used the recommended best mode (most time domain - Lambda1) for that kind with highest possible quality. It was (compared to the frequency domain calculation) relatively fast, but much too slow for realtime, I think.
speech 50% time stretched
speech original
speech 200% time stretched
2. A music sample.
I used the recommended best mode (most frequency domain - Lambda5) with highest possible quality (see next comment).
Notes:
- I got crashes, when using frequency domain (Lambda6) which were caused inside the
binary library (so out of my focus).
- Frequency domain processing needed extraordinary long time (over 5 minutes!!!) to process on those small files.
music 50% time stretched
music original
music 200% time stretched
No. I finally don't see the hype.
(But I do NOT say, that it is bad.)
-
- KVRian
- Topic Starter
- 568 posts since 20 Nov, 2003 from Basel, Switzerland
What bullshit are you referring to?? I wasn't talking bullshit - I just questioned your attitude.
I didn't say there was a hype. I just said it's the best time stretching *I* have ever heard. 50% and 200% hardly says anything about quality. Usually you need from 80-130% for most uses on mixes and tracks. I had no crashes so far (xcode on tha Mac, as I said) and it's *very* fast (except when you use the highest quality setting which I think you don't need for most signals anyway).
I'm getting the best results with the pre-set quality and lambda values which, judging from the speed of the processing, is several times realtime speed on my G5...
And as far as I am concerned, offline processing at maximum quality is fine with me. I've never heard a time stretch that produced usable quality at a realtime setting anyway, and computers are getting faster each day...
--th
I didn't say there was a hype. I just said it's the best time stretching *I* have ever heard. 50% and 200% hardly says anything about quality. Usually you need from 80-130% for most uses on mixes and tracks. I had no crashes so far (xcode on tha Mac, as I said) and it's *very* fast (except when you use the highest quality setting which I think you don't need for most signals anyway).
I'm getting the best results with the pre-set quality and lambda values which, judging from the speed of the processing, is several times realtime speed on my G5...
And as far as I am concerned, offline processing at maximum quality is fine with me. I've never heard a time stretch that produced usable quality at a realtime setting anyway, and computers are getting faster each day...
--th
I'm the stereo chancellor
-
- KVRian
- Topic Starter
- 568 posts since 20 Nov, 2003 from Basel, Switzerland
Hallo Günter,
Maybe you should listen to some of those tracks: http://www.kuhn-management.de which is where, according to "whois dizainer.net", you appear to be working (I'll go vomit now...)
--th
About having no clue: are you're attacking me now, or is that just your preferred tone for conversation?jackle&hyde wrote:[Bla bla bla... (You obviously have no clue, what you are talking about).
Maybe you should listen to some of those tracks: http://www.kuhn-management.de which is where, according to "whois dizainer.net", you appear to be working (I'll go vomit now...)
--th
I'm the stereo chancellor
-
- KVRist
- 240 posts since 19 Mar, 2004 from London UK
quite good! would be great if formants could be preserved for pitch shifting
-
- KVRAF
- 25042 posts since 12 Jul, 2003 from West Caprazumia
dunno what this got to do with the topic of this thread...tahome wrote: Maybe you should listen to some of those tracks: http://www.kuhn-management.de which is where, according to "whois dizainer.net", you appear to be working (I'll go vomit now...)
--th
(or in other words: 'what is your point?')
-
- KVRian
- Topic Starter
- 568 posts since 20 Nov, 2003 from Basel, Switzerland
Well, my point is that 1) I do have a clue, 2) someone who listens to this kind of music is hardly qualified for a judgement about quality and 3) I can understand why someone who has to listen to Schlager music all day is inherently violent....
Now don't take me too seriously, ok?
--th
Now don't take me too seriously, ok?
--th
I'm the stereo chancellor
-
- KVRAF
- 8389 posts since 11 Apr, 2003 from back on the hillside again - but now with a garden!
please girls, handbags down!
both of you have views, and I think some crossed wires. I don't think there's a war here really, just some misunderstandings about explanations.
Dirac looks intersting, but won't be for everyone. It'll be interesting to see what gets coded, and regarding VST and sample buffers, etc, it doesn't really matter if you internally handle a buffer of different length to the input - otherwise delays and reverbs wouldn't work.. You just initialise a buffer that you think should be long enough, add 5% for good measure, let dirac run for the buffer VST hands you, and crank out where you've got upto into the o/p buffer.. etc. You may need to do some SOLA to avoid glitches, and there will be some latency issues..
but it's not all bad!
I won't be using it, having looked it over. You can get around the single channel/fixed sample rate thing but doing your own samplerate conversions, and employing extra objects, etc. But even so, I would prefer to use my own algo's and have a laugh learning! SMB has already given a good pitch scale algo for free, it's a case of adapt and learn.
DSP
both of you have views, and I think some crossed wires. I don't think there's a war here really, just some misunderstandings about explanations.
Dirac looks intersting, but won't be for everyone. It'll be interesting to see what gets coded, and regarding VST and sample buffers, etc, it doesn't really matter if you internally handle a buffer of different length to the input - otherwise delays and reverbs wouldn't work.. You just initialise a buffer that you think should be long enough, add 5% for good measure, let dirac run for the buffer VST hands you, and crank out where you've got upto into the o/p buffer.. etc. You may need to do some SOLA to avoid glitches, and there will be some latency issues..
but it's not all bad!
I won't be using it, having looked it over. You can get around the single channel/fixed sample rate thing but doing your own samplerate conversions, and employing extra objects, etc. But even so, I would prefer to use my own algo's and have a laugh learning! SMB has already given a good pitch scale algo for free, it's a case of adapt and learn.
DSP
-
- KVRian
- Topic Starter
- 568 posts since 20 Nov, 2003 from Basel, Switzerland
I'm not at all interested in starting a war here... but I just can't stand people who can only see the negative side of things. Anyway, as I said before, I'm amazed by the quality and I'd very much like to see a product that uses it. If it's commercial that's fine with me - I'd buy it rightaway. I can't develop it myself and that's why I brought it up here, and I'd be happy to get involved in a constructive discussion about it...
If this isn't possible or not welcome I'd be happy to post my plea someplace else.
--th
If this isn't possible or not welcome I'd be happy to post my plea someplace else.
--th
I'm the stereo chancellor
-
- KVRian
- 1325 posts since 1 Sep, 2004
Did you mean the test server, I posted the samples?tahome wrote:Hallo Günter,
About having no clue: are you're attacking me now, or is that just your preferred tone for conversation?jackle&hyde wrote:[Bla bla bla... (You obviously have no clue, what you are talking about).
Maybe you should listen to some of those tracks: http://www.kuhn-management.de which is where, according to "whois dizainer.net", you appear to be working (I'll go vomit now...)
--th
(What a "wannabe" smarty you are - did you try a whois?)
Well, Dude, that's a public server for even such things....
You should be somehow more carefull, to "involve" people, who really have nothing to do with that...
Kuhn-Management is (to your information) an artist management company in Germany, Switzerland and Austria...
And: No. My name is not Günter.
BUT: Why do I post on those servers?
Maybe because I'm an artist? Or a producer? or a music studio men? Or I am an entire music company? Or simply a men who has got authorized access to a subdomain of a public server?
Last edited by useruseruser on Wed May 25, 2005 1:20 pm, edited 1 time in total.
-
- KVRian
- 1325 posts since 1 Sep, 2004
Listen to that samples I posted above. They are constuctive enuff.
You hateful dude, (not being able to do a "normal" discussion and not able to live with other peoples opinion.)
Probably you are even one of those guys, who spend too much time infront of their computer, loosing any ability to solve "real live" conflicts in any normal fashion.
You hateful dude, (not being able to do a "normal" discussion and not able to live with other peoples opinion.)
Probably you are even one of those guys, who spend too much time infront of their computer, loosing any ability to solve "real live" conflicts in any normal fashion.
Last edited by useruseruser on Wed May 25, 2005 1:20 pm, edited 1 time in total.