[Csnd] problem when opening 10 stereo files
Date | 2011-12-20 17:17 |
From | Michele Nasti |
Subject | [Csnd] problem when opening 10 stereo files |
Hi there, I am realizing a nice program that needs to play 10 ~ 20 stereo files all at once. I thought csound could do this, infact I am doing this :) since I have to do some time scaling on it, I use pvstanal or mincer to stretch it ... It takes a mono source. So for every file loaded (.aif or .ogg) I create 2 function tables, one per channel, and then I unify them with the outch opcode.
The result is a bit crappy, I hear that something is lost, and it scratches. Is there any idea to get more memory, or to make it work in a more fluid way? Any suggestions or ideas? Thanks in advance
Michele
|
Date | 2011-12-20 19:34 |
From | peiman khosravi |
Subject | Re: [Csnd] problem when opening 10 stereo files |
Try increasing the buffer size (with the -b and -B flags) decreasing ksmp size. But I have to say that having 20-40 instances of pvstanal is moving towards an impossibility. P On 20 December 2011 17:17, Michele Nasti |
Date | 2011-12-20 19:51 |
From | Victor Lazzarini |
Subject | Re: [Csnd] problem when opening 10 stereo files |
Also check your CPU, to see if it's maxing out. On 20 Dec 2011, at 19:34, peiman khosravi wrote: > Try increasing the buffer size (with the -b and -B flags) decreasing ksmp size. > > But I have to say that having 20-40 instances of pvstanal is moving > towards an impossibility. > > P > > On 20 December 2011 17:17, Michele Nasti |
Date | 2011-12-20 19:52 |
From | peiman khosravi |
Subject | Re: [Csnd] problem when opening 10 stereo files |
Sorry I meant increasing ksmp. On 20 December 2011 19:34, peiman khosravi |
Date | 2011-12-21 00:39 |
From | Michele Nasti |
Subject | Re: [Csnd] problem when opening 10 stereo files |
I tried these suggestions but the problem is too big. Another problem I found is that I have 1 GB of audio function tables loaded in memory :) I'm asking myself how the various Cubase, Logic etc can reproduce 20-30 stereo files, with live effects and without this cpu effort. It's not criticism, I know that csound and these softwares have 2 different objectives, but I'm curious to know techically how they are realized.
I think I'm going to try two optimizations: 1. move files from stereo to mono (extreme solution, the cpu power needed should became half), 2. change the sound quality (from 44100 sr to 22050).
Based on your experience, can you tell me if the second option is praticable? Michele 2011/12/20 peiman khosravi <peimankhosravi@gmail.com> Sorry I meant increasing ksmp. |
Date | 2011-12-21 01:52 |
From | peiman khosravi |
Subject | Re: [Csnd] problem when opening 10 stereo files |
I believe that DAWs read the audio directly from disk, with a limited buffer size. You'd probably be OK reading 20 stereo files with diskin. Also these are multithreading. The trouble here is not only your tables but also the PVS analysis which is heavy duty. You'd also find it impossible to run, say, 20 instances of GRMtools spectral plug-ins in Protools! Imagine what I am going through with a 6 channel composition at 96k. Complete nightmare! P On 21 December 2011 00:39, Michele Nasti |
Date | 2011-12-21 03:44 |
From | peiman khosravi |
Subject | Re: [Csnd] problem when opening 10 stereo files |
On 21 December 2011 00:39, Michele Nasti |
Date | 2011-12-21 09:58 |
From | Tito Latini |
Subject | Re: [Csnd] problem when opening 10 stereo files |
Attachments | None |
Date | 2011-12-21 10:16 |
From | Victor Lazzarini |
Subject | Re: [Csnd] problem when opening 10 stereo files |
Definitely an alternative. Also you can mix them down with pvsmix so to save on FFTs for resynthesis. If you mix by pairs, then you only need half of the pvsynth instances. Finally, just note that the timestretching quality will be different as pvsfread involves a different method. For small changes, there may not be much difference. Victor However, the timestretching will sound a bit different. On 21 Dec 2011, at 09:58, Tito Latini wrote: > A possible solution is to use the pvx files generated > through the `pvanal' utility. In this way you avoid > the gens-01 and skip the pvs analysis in csound. > You cannot use `mincer' and `pvstanal' anymore, but > > fsig1 pvsfread ktimpt, "pvxfile", 0 ;; chan 1 > fsig2 pvsfread ktimpt, "pvxfile", 1 ;; chan 2 > ;; optionally whatever pvs opcode that uses a fsig (also see pvslock) > ... > al pvsynth fsigout1 > ar pvsynth fsigout2 > > tito > > On Wed, Dec 21, 2011 at 01:39:48AM +0100, Michele Nasti wrote: >> I tried these suggestions but the problem is too big. >> >> Another problem I found is that I have 1 GB of audio function tables loaded >> in memory :) >> >> I'm asking myself how the various Cubase, Logic etc can reproduce 20-30 >> stereo files, with live effects and without this cpu effort. It's not >> criticism, I know that csound and these softwares have 2 different >> objectives, but I'm curious to know techically how they are realized. >> >> I think I'm going to try two optimizations: 1. move files from stereo to >> mono (extreme solution, the cpu power needed should became half), 2. change >> the sound quality (from 44100 sr to 22050). > > > Send bugs reports to the Sourceforge bug tracker > https://sourceforge.net/tracker/?group_id=81968&atid=564599 > Discussions of bugs and features can be posted here > To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound" > Dr Victor Lazzarini Senior Lecturer Dept. of Music NUI Maynooth Ireland tel.: +353 1 708 3545 Victor dot Lazzarini AT nuim dot ie Send bugs reports to the Sourceforge bug tracker https://sourceforge.net/tracker/?group_id=81968&atid=564599 Discussions of bugs and features can be posted here To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound" |