[Csnd] funny randomi issue
Date | 2011-01-25 20:39 |
From | peiman khosravi |
Subject | [Csnd] funny randomi issue |
randomi seems to get stuck on zero for a short while at the start of the performance. Thanks Peiman |
Date | 2011-01-26 08:35 |
From | menno |
Subject | [Csnd] Re: funny randomi issue |
Hi Peiman, FYI it does not get stuck here. I use Csound5.13beta, doubles ,Ubuntu, 48kHz, Jack: new alloc for instr 1: i1 0.00000 i1 0.00052 i1 0.00111 i1 0.00170 i1 0.00229 i1 0.00288 i1 0.00347 i1 0.00406 i1 0.00464 i1 0.00523 i1 0.00582 i1 0.00641 i1 0.00700 i1 0.00759..... greetings menno |
Date | 2011-01-26 09:21 |
From | peiman khosravi |
Subject | Re: [Csnd] Re: funny randomi issue |
Thanks Menno, I am on OS X 10.6. Using this other csd bellow I get the following. The value remains on zero for .5 seconds every time. new alloc for instr 1: i 1 time 0.06250: 0.00000 i 1 time 0.12500: 0.00000 i 1 time 0.25000: 0.00000 i 1 time 0.31250: 0.00000 i 1 time 0.43750: 0.00000 i 1 time 0.50000: 0.00000 i 1 time 0.62500: 0.08407 i 1 time 0.75000: 0.25222 i 1 time 0.81250: 0.33630 i 1 time 0.93750: 0.50445 i 1 time 1.00000: 0.58852 i 1 time 1.12500: 0.69159 i 1 time 1.25000: 0.72958 i 1 time 1.31250: 0.74858 i 1 time 1.43750: 0.78657 i 1 time 1.50000: 0.80556 i 1 time 1.62500: 0.78747 i 1 time 1.75000: 0.71329 Best, Peiman |
Date | 2011-01-26 10:36 |
From | peiman khosravi |
Subject | Re: [Csnd] Re: funny randomi issue |
I wonder if this is only happening with this opcode or if it's a more general OS X related issue to do with krate and timing. P On 26 January 2011 09:21, peiman khosravi |
Date | 2011-01-26 22:05 |
From | joachim heintz |
Subject | Re: [Csnd] Re: funny randomi issue |
do you also use 5.13? Am 26.01.2011 11:36, schrieb peiman khosravi: > I wonder if this is only happening with this opcode or if it's a more > general OS X related issue to do with krate and timing. > > P > > On 26 January 2011 09:21, peiman khosravi |
Date | 2011-01-26 22:08 |
From | peiman khosravi |
Subject | Re: [Csnd] Re: funny randomi issue |
yes 5.13. I used the universal installer. On 26 January 2011 22:05, joachim heintz |
Date | 2011-01-26 22:15 |
From | Victor Lazzarini |
Subject | Re: [Csnd] Re: funny randomi issue |
Not stuck here. 5.13 OSX 10.5 i1 0.00000 i1 0.11036 i1 0.22072 i1 0.33108 i1 0.44144 i1 0.55180 i1 0.66215 i1 0.77251 i1 0.88287 i1 0.80893 i1 0.73499 i1 0.66105 i1 0.58711 i1 0.51317 i1 0.43922 i1 0.36528 i1 0.29134 i1 0.37064 i1 0.44993 i1 0.52923 i1 0.60853 i1 0.68782 i1 0.76712 i1 0.84641 i1 0.92571 i1 0.82436 i1 0.72301 i1 0.62166 i1 0.52031 i1 0.41896 i1 0.31761 i1 0.21626 i1 0.11491 i1 0.10123 i1 0.08756 i1 0.07388 i1 0.06021 i1 0.04653 i1 0.03286 i1 0.01918 i1 0.00551 i1 0.06780 i1 0.13008 i1 0.19237 i1 0.25466 i1 0.31695 i1 0.37924 i1 0.44153 i1 0.50381 i1 0.51766 i1 0.53150 i1 0.54534 i1 0.55919 i1 0.57303 i1 0.58687 i1 0.60072 On 26 Jan 2011, at 22:08, peiman khosravi wrote: > yes 5.13. I used the universal installer. > > > > On 26 January 2011 22:05, joachim heintz |
Date | 2011-01-26 23:03 |
From | peiman khosravi |
Subject | Re: [Csnd] Re: funny randomi issue |
Can you try this one please Victor. |
Date | 2011-01-27 00:56 |
From | peiman khosravi |
Subject | Re: [Csnd] Re: funny randomi issue |
Hi Victor, Just to clarify. What I mean by 'stuck' is that there is a .5 seconds delay between the first value that is always zero and the second value. The original csd I sent is not very good for debuging actually as it uses printk2. The delay between the first and second value is obvious when you use printk that prints out the time as well. I am getting the same issue on my 10.5 machine. Best, Peiman On 26 January 2011 23:03, peiman khosravi |
Date | 2011-01-27 08:00 |
From | Victor Lazzarini |
Subject | Re: [Csnd] Re: funny randomi issue |
yes, with the second CSD, I get the repeated 0s. On 27 Jan 2011, at 00:56, peiman khosravi wrote: > Hi Victor, > > Just to clarify. What I mean by 'stuck' is that there is a .5 seconds > delay between the first value that is always zero and the second > value. > > The original csd I sent is not very good for debuging actually as it > uses printk2. The delay between the first and second value is obvious > when you use printk that prints out the time as well. > > I am getting the same issue on my 10.5 machine. > > Best, > > Peiman > > On 26 January 2011 23:03, peiman khosravi |
Date | 2011-01-27 08:18 |
From | Francois PINOT |
Subject | Re: [Csnd] Re: funny randomi issue |
Hi Peiman, This is normal behavior: randomi is an interpolating opcode. That means that it will generate new random numbers at a rate of kcps but, during the 1/kcps sec separating two random numbers, it will output interpolated values. The first values in your example are always 0 because the first interpolation is done between kmin (which is 0 in your example) and the first generated random value. By changing the value of kmin, you'll get different starting values. Moreover, if you change the kcps value, the number of fixed initial values will change accordingly. Hope this helps Francois 2011/1/27 peiman khosravi <peimankhosravi@gmail.com> Can you try this one please Victor. |
Date | 2011-01-27 09:02 |
From | peiman khosravi |
Subject | Re: [Csnd] Re: funny randomi issue |
Thanks for the explaination Francois. So just to clarify. The minimum value is always outputed for 1/kcps seconds when the note is initialised? Is this because the opcode uses this time to generate the next random number that is subsequently used for the interpolation? I think this should be in the manual as a kcps value of 0.1 or somthing like that means a long wait (10 seconds!). I guess the trick would be something like this: kmult linseg 1000, 0.001, 1, p3, 1 kmorph2 randomi 0,1, 2*kmult Best, Peiman On 27 January 2011 08:18, Francois PINOT |
Date | 2011-01-27 10:42 |
From | Francois PINOT |
Subject | Re: [Csnd] Re: funny randomi issue |
Absolutely! I can add this paragraph to the "Performance" section of the manual entry for randomi: The kmin value is outputted for 1/kcps (resp. 1/acps) seconds at the beginning of the note, before the first random number is generated. Then the normal interpolation process takes place, first between kmin and the first random number generated, and then between successive generated random numbers, each interpolation cycle having a duration of 1/kcps (resp. 1/acps) seconds. Is it OK? Francois 2011/1/27 peiman khosravi <peimankhosravi@gmail.com> Thanks for the explaination Francois. |
Date | 2011-01-27 10:56 |
From | Andres Cabrera |
Subject | Re: [Csnd] Re: funny randomi issue |
It might actually be a better solution to add an optional i-time parameter which sets the initial state, or an option to enable producing a random number from the start. Cheers, Andres On Thu, Jan 27, 2011 at 10:42 AM, Francois PINOT |
Date | 2011-01-27 11:10 |
From | peiman khosravi |
Subject | Re: [Csnd] Re: funny randomi issue |
Thanks! Yes I second Andres. I think the addition of both, an initial state and starting the interpolation immediately would be a big bonus. Best, Peiman On 27 January 2011 10:56, Andres Cabrera |
Date | 2011-01-27 14:49 |
From | Francois PINOT |
Subject | Re: [Csnd] funny randomi issue |
Something like this? kres randomi kmin, kmax, kcps [,iinterp0] [,ifirstval] iinterp0 (optional, default=0) -- first interpolation cycle mode (see below) ifirstval (optional, default=0) -- first output value When iinterp0 = 0 (the default), the kmin argument value is outputted for 1/kcps (resp. 1/acps) seconds at the beginning of the note, before the first random number is generated. Then the normal interpolation process takes place, first between kmin and the first random number generated, and then between successive generated random numbers, each interpolation cycle having a duration of 1/kcps (resp. 1/acps) seconds. When iinterp0 = 1, a random number is generated at initialization and interpolation begins immediately between the kmin argument value and that random number. When interp0 = 2, a random number is generated at initialization and interpolation begins immediately between the ifirstval argument value and that random number. This way, we could have immediate interpolation and initial value without modifying older versions behavior. If you agree with this, I can modify the source files (opcodes/uggab.h and opcodes/uggab.c), test the modifications and send the files to Andrés for upload Regards François 2011/1/27 peiman khosravi <peimankhosravi@gmail.com> Thanks! |
Date | 2011-01-27 15:57 |
From | Andres Cabrera |
Subject | Re: [Csnd] funny randomi issue |
Thanks Francois, that looks great. Cheers, Andres On Thu, Jan 27, 2011 at 2:49 PM, Francois PINOT |