Hi,
attached is some stretched phasor~ fun for your classic synthesis methods class: an implementation of the VOSIM algorithm for speech synthesis etc. as invented by Kaegi/Templaars.
My implementation is almost correct, but has a slight bug, which you can see if you set M and T to the values indicated by message boxes, set N to 2 or 3 and and set b to be larger than 0. Then the step function which scales the amplitude of the sin^2 pulses starts irregularily. I'm quite sure this comes from the way I use [samphold~] inside [vosim~] as it sometimes samples the wrong value. However I don't yet know of a proper workaround. (One workaround would be to just omit the samhold-step, but the result would not be VOSIM according to the textbooks.)
So anybody with a good idea?
Frank Barknecht _ ______footils.org__
I see the occasional spike on the rhs of samplehold so rescaling and clipping seems to fix it. Not a good solution because it fixes the symptom not the cause, but seems to work.
On Tue, 11 Dec 2007 15:07:17 +0100 Frank Barknecht fbar@footils.org wrote:
Hi,
attached is some stretched phasor~ fun for your classic synthesis methods class: an implementation of the VOSIM algorithm for speech synthesis etc. as invented by Kaegi/Templaars.
My implementation is almost correct, but has a slight bug, which you can see if you set M and T to the values indicated by message boxes, set N to 2 or 3 and and set b to be larger than 0. Then the step function which scales the amplitude of the sin^2 pulses starts irregularily. I'm quite sure this comes from the way I use [samphold~] inside [vosim~] as it sometimes samples the wrong value. However I don't yet know of a proper workaround. (One workaround would be to just omit the samhold-step, but the result would not be VOSIM according to the textbooks.)
So anybody with a good idea?
Ciao
Frank Barknecht _ ______footils.org__
Hmm, isn't the second [wrap~] redundant anyway?
On Tue, 11 Dec 2007 15:07:17 +0100 Frank Barknecht fbar@footils.org wrote:
Hi,
attached is some stretched phasor~ fun for your classic synthesis methods class: an implementation of the VOSIM algorithm for speech synthesis etc. as invented by Kaegi/Templaars.
My implementation is almost correct, but has a slight bug, which you can see if you set M and T to the values indicated by message boxes, set N to 2 or 3 and and set b to be larger than 0. Then the step function which scales the amplitude of the sin^2 pulses starts irregularily. I'm quite sure this comes from the way I use [samphold~] inside [vosim~] as it sometimes samples the wrong value. However I don't yet know of a proper workaround. (One workaround would be to just omit the samhold-step, but the result would not be VOSIM according to the textbooks.)
So anybody with a good idea?
Ciao
Frank Barknecht _ ______footils.org__
Hallo, Andy Farnell hat gesagt: // Andy Farnell wrote:
Hmm, isn't the second [wrap~] redundant anyway?
Theoretically yes, but practically it seems that adding the second wrap~ after wrap~ is exactly what fixes the problem! I even could omit the additions you made by just doing the seemingly idiotic:
| [wrap~] | [wrap~] |
(see attached)
I'm a bit stumped!
Frank Barknecht _ ______footils.org__
Hmm, isn't the second [wrap~] redundant anyway?
Theoretically yes, but practically it seems that adding the second wrap~ after wrap~ is exactly what fixes the problem!
http://lists.puredata.info/pipermail/pd-list/2007-05/049886.html
Hope this helps...
Claude
Hallo, Claude Heiland-Allen hat gesagt: // Claude Heiland-Allen wrote:
Hmm, isn't the second [wrap~] redundant anyway?
Theoretically yes, but practically it seems that adding the second wrap~ after wrap~ is exactly what fixes the problem!
http://lists.puredata.info/pipermail/pd-list/2007-05/049886.html
Hope this helps...
Ah, it's that one. Okay, I'll add one then.
Frank Barknecht _ ______footils.org__
Hi,
actually wrap~ seems really buggy. Maybe its the nature of its concept, anyway my fix is to wrap wrap~ into [wrap-switch~] that has a sig~ connected to the wrap~ inlet and then a [r wrapswitch]. Now you can change the wrapping on all wrap~´s just by sending 0,1 or even 2 to [r wrapswitch].
Maybe I´m just beeing neurotic but there is some kind of very incosistent bug, maybe some kind of overflow. I´ll try once more to track it.
mvh/"the other" Steffen (Leve Poulsen) on XP Pd version 0.40-2
Frank Barknecht skrev:
Hallo, Claude Heiland-Allen hat gesagt: // Claude Heiland-Allen wrote:
Hmm, isn't the second [wrap~] redundant anyway?
Theoretically yes, but practically it seems that adding the second wrap~ after wrap~ is exactly what fixes the problem!
http://lists.puredata.info/pipermail/pd-list/2007-05/049886.html
Hope this helps...
Ah, it's that one. Okay, I'll add one then.
Ciao
is this just me, or why are Andys emails always in the future ? my date is atm: Tue Dec 11 19:03:36 CET 2007
.andre
On Wed, 2007-12-12 at 04:21 +0000, Andy Farnell wrote:
Hmm, isn't the second [wrap~] redundant anyway?
On Tue, 11 Dec 2007 15:07:17 +0100 Frank Barknecht fbar@footils.org wrote:
Hi,
attached is some stretched phasor~ fun for your classic synthesis methods class: an implementation of the VOSIM algorithm for speech synthesis etc. as invented by Kaegi/Templaars.
My implementation is almost correct, but has a slight bug, which you can see if you set M and T to the values indicated by message boxes, set N to 2 or 3 and and set b to be larger than 0. Then the step function which scales the amplitude of the sin^2 pulses starts irregularily. I'm quite sure this comes from the way I use [samphold~] inside [vosim~] as it sometimes samples the wrong value. However I don't yet know of a proper workaround. (One workaround would be to just omit the samhold-step, but the result would not be VOSIM according to the textbooks.)
So anybody with a good idea?
Ciao
Frank Barknecht _ ______footils.org__
no, here as well. and afaik, it's only andy's mails. probably not having set the clock to utc?
roman
On Tue, 2007-12-11 at 19:04 +0100, Andre Schmidt wrote:
is this just me, or why are Andys emails always in the future ? my date is atm: Tue Dec 11 19:03:36 CET 2007
.andre
On Wed, 2007-12-12 at 04:21 +0000, Andy Farnell wrote:
Hmm, isn't the second [wrap~] redundant anyway?
On Tue, 11 Dec 2007 15:07:17 +0100 Frank Barknecht fbar@footils.org wrote:
Hi,
attached is some stretched phasor~ fun for your classic synthesis methods class: an implementation of the VOSIM algorithm for speech synthesis etc. as invented by Kaegi/Templaars.
My implementation is almost correct, but has a slight bug, which you can see if you set M and T to the values indicated by message boxes, set N to 2 or 3 and and set b to be larger than 0. Then the step function which scales the amplitude of the sin^2 pulses starts irregularily. I'm quite sure this comes from the way I use [samphold~] inside [vosim~] as it sometimes samples the wrong value. However I don't yet know of a proper workaround. (One workaround would be to just omit the samhold-step, but the result would not be VOSIM according to the textbooks.)
So anybody with a good idea?
Ciao
Frank Barknecht _ ______footils.org__
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Telefonate ohne weitere Kosten vom PC zum PC: http://messenger.yahoo.de
would it be possible to get the timestamp from the mailinglist server? marius.
Roman Haefeli wrote:
no, here as well. and afaik, it's only andy's mails. probably not having set the clock to utc?
roman
On Tue, 2007-12-11 at 19:04 +0100, Andre Schmidt wrote:
is this just me, or why are Andys emails always in the future ? my date is atm: Tue Dec 11 19:03:36 CET 2007
.andre
On Wed, 2007-12-12 at 04:21 +0000, Andy Farnell wrote:
Hmm, isn't the second [wrap~] redundant anyway?
On Tue, 11 Dec 2007 15:07:17 +0100 Frank Barknecht fbar@footils.org wrote:
Hi,
attached is some stretched phasor~ fun for your classic synthesis methods class: an implementation of the VOSIM algorithm for speech synthesis etc. as invented by Kaegi/Templaars.
My implementation is almost correct, but has a slight bug, which you can see if you set M and T to the values indicated by message boxes, set N to 2 or 3 and and set b to be larger than 0. Then the step function which scales the amplitude of the sin^2 pulses starts irregularily. I'm quite sure this comes from the way I use [samphold~] inside [vosim~] as it sometimes samples the wrong value. However I don't yet know of a proper workaround. (One workaround would be to just omit the samhold-step, but the result would not be VOSIM according to the textbooks.)
So anybody with a good idea?
Ciao
Frank Barknecht _ ______footils.org__
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
___________________________________________________________ Telefonate ohne weitere Kosten vom PC zum PC: http://messenger.yahoo.de
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
marius schebella wrote:
would it be possible to get the timestamp from the mailinglist server?
you already do. just have a look at the header of any(!) standard-conformant email (and i am happy to say that at least the headers added by the pd mailing list server are conformant to various RFCs)
mfg.asdr IOhannes
Yep, Claude picked up on this and I changed it. Then it changed itself back.
So someone else mentioned it and next time I reset the motherboard. It changed itself back.
I put ntpd on. When it boots it has the right time for a while and then it ignores that and goes back to the future.
I just changed the ntp settings again to a new server and update rate.
If this doesn't work, how do you go about getting a computer exorcism?
On Tue, 11 Dec 2007 19:04:14 +0100 Andre Schmidt andre@osku.de wrote:
is this just me, or why are Andys emails always in the future ? my date is atm: Tue Dec 11 19:03:36 CET 2007
.andre
On Wed, 2007-12-12 at 04:21 +0000, Andy Farnell wrote:
Hmm, isn't the second [wrap~] redundant anyway?
On Tue, 11 Dec 2007 15:07:17 +0100 Frank Barknecht fbar@footils.org wrote:
Hi,
attached is some stretched phasor~ fun for your classic synthesis methods class: an implementation of the VOSIM algorithm for speech synthesis etc. as invented by Kaegi/Templaars.
My implementation is almost correct, but has a slight bug, which you can see if you set M and T to the values indicated by message boxes, set N to 2 or 3 and and set b to be larger than 0. Then the step function which scales the amplitude of the sin^2 pulses starts irregularily. I'm quite sure this comes from the way I use [samphold~] inside [vosim~] as it sometimes samples the wrong value. However I don't yet know of a proper workaround. (One workaround would be to just omit the samhold-step, but the result would not be VOSIM according to the textbooks.)
So anybody with a good idea?
Ciao
Frank Barknecht _ ______footils.org__
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
On 11/12/2007, at 7:56 PM, Andy Farnell wrote:
I put ntpd on. When it boots it has the right time for a while and then it ignores that and goes back to the future.
"Back to the future"....a suspicious effect
If this doesn't work, how do you go about getting a computer exorcism?
Having Michael J Fox exorcised is hard. As we know, he is one evil motherfucker, as Mojo Nixon and Skid Roper have told us:
"Elvis is in everybody out there. Everybody's got Elvis in them. Everybody except one person that is, Yeah, one person! The evil opposite of Elvis, The Anti-Elvis! Anti-Elvis got no Elvis in him, Lemme tell ya.
Michael J. Fox has no Elvis in him."
http://www.geocities.com/multielvi/mojo.html
I wish you luck. Perhaps on Elvis' birthday, 8 Jan, you can perform a sacrificial burning of a VHS copy of "Back to the future" and send that evil spirit away......
cool, thanks frank. this makes a nice squelchy type of synth if you slide the values.