Hello List,
I have some Gem patches that react to incoming sound in various ways.
Up until now, I would take the adc~ input and use 3 vcf~ 's (one for
bass, mid and high) and with the right settings of Hz and Q, I was
getting fairly good envelopes.
Looking at this again and wanting to improve it, I have been
experimenting with iemlib filters:
http://pd.iem.at/iemlib/abstract.html
I was wondering if anyone can recommend which of these would be best
for trying to isolate, say a bass drum or a hi hat from an incoming mix?
I read the Bessel filters are often used in cross overs but from just
experimenting a bit it seems some of the others (e.g. vcf_lp2~) are
giving me a better response.
Thanks, p
Instead of filters, try using bonk~, which you can train tto differentiate various attacks and label them accordingly in its output.
~Kyle
On 9/1/06, Paris Treantafeles paris@parisgraphics.com wrote:
Hello List,
I have some Gem patches that react to incoming sound in various ways. Up until now, I would take the adc~ input and use 3 vcf~ 's (one for bass, mid and high) and with the right settings of Hz and Q, I was getting fairly good envelopes. Looking at this again and wanting to improve it, I have been experimenting with iemlib filters: http://pd.iem.at/iemlib/abstract.html
I was wondering if anyone can recommend which of these would be best for trying to isolate, say a bass drum or a hi hat from an incoming mix?
I read the Bessel filters are often used in cross overs but from just experimenting a bit it seems some of the others (e.g. vcf_lp2~) are giving me a better response.
Thanks, p
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Hi Kyle,
Thanks - I should have mentioned that I use bonk~ in some patches as
well.
I've had better luck with it though when there's not too many
different simultaneous frequencies going on.
That's what got me filtering the input first- but I will take another
look at it, maybe i just didn't use it's full possibilities.
-p
On Sep 1, 2006, at 3:41 PM, Kyle Klipowicz wrote:
Instead of filters, try using bonk~, which you can train tto differentiate various attacks and label them accordingly in its output.
~Kyle
On 9/1/06, Paris Treantafeles paris@parisgraphics.com wrote:
Hello List,
I have some Gem patches that react to incoming sound in various ways. Up until now, I would take the adc~ input and use 3 vcf~ 's (one for bass, mid and high) and with the right settings of Hz and Q, I was getting fairly good envelopes. Looking at this again and wanting to improve it, I have been experimenting with iemlib filters: http://pd.iem.at/iemlib/abstract.html
I was wondering if anyone can recommend which of these would be best for trying to isolate, say a bass drum or a hi hat from an
incoming mix?I read the Bessel filters are often used in cross overs but from just experimenting a bit it seems some of the others (e.g. vcf_lp2~) are giving me a better response.
Thanks, p
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
--
http://theradioproject.com http://perhapsidid.blogspot.com
(((())))(()()((((((((()())))()(((((((())()()())()))) (())))))(()))))))))))))(((((((((((()()))))))))((()))) ))(((((((((((())))())))))))))))))))__________ _____())))))(((((((((((((()))))))))))_______ ((((((())))))))))))((((((((000)))oOOOOOO
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
Paris~
I really would recommend the "learn" method, it creates a fft-based profile (I think using a neural net, correct me someone if I'm mistaken) of the sounds through a training routine of a suggested 10 occurrences.
I have not played with this feature too much myself, but imagine that if it works even halfway decent, it could do much to better your applications. Please keep the list updated on your success/failure with this feature.
~Kyle
On 9/1/06, Paris Treantafeles paris@parisgraphics.com wrote:
Hi Kyle,
Thanks - I should have mentioned that I use bonk~ in some patches as well. I've had better luck with it though when there's not too many different simultaneous frequencies going on. That's what got me filtering the input first- but I will take another look at it, maybe i just didn't use it's full possibilities.
-p
On Sep 1, 2006, at 3:41 PM, Kyle Klipowicz wrote:
Instead of filters, try using bonk~, which you can train tto differentiate various attacks and label them accordingly in its output.
~Kyle
On 9/1/06, Paris Treantafeles paris@parisgraphics.com wrote:
Hello List,
I have some Gem patches that react to incoming sound in various ways. Up until now, I would take the adc~ input and use 3 vcf~ 's (one for bass, mid and high) and with the right settings of Hz and Q, I was getting fairly good envelopes. Looking at this again and wanting to improve it, I have been experimenting with iemlib filters: http://pd.iem.at/iemlib/abstract.html
I was wondering if anyone can recommend which of these would be best for trying to isolate, say a bass drum or a hi hat from an incoming mix?
I read the Bessel filters are often used in cross overs but from just experimenting a bit it seems some of the others (e.g. vcf_lp2~) are giving me a better response.
Thanks, p
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
--
http://theradioproject.com http://perhapsidid.blogspot.com
(((())))(()()((((((((()())))()(((((((())()()())()))) (())))))(()))))))))))))(((((((((((()()))))))))((()))) ))(((((((((((())))())))))))))))))))__________ _____())))))(((((((((((((()))))))))))_______ ((((((())))))))))))((((((((000)))oOOOOOO
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
Hi Kyle,
Good call - that's exactly what I didn't spend much time on! I'll have to read up and experiment with that.
Thanks, p
On Sep 1, 2006, at 3:56 PM, Kyle Klipowicz wrote:
Paris~
I really would recommend the "learn" method, it creates a fft-based profile (I think using a neural net, correct me someone if I'm mistaken) of the sounds through a training routine of a suggested 10 occurrences.
I have not played with this feature too much myself, but imagine that if it works even halfway decent, it could do much to better your applications. Please keep the list updated on your success/failure with this feature.
~Kyle
On 9/1/06, Paris Treantafeles paris@parisgraphics.com wrote:
Hi Kyle,
Thanks - I should have mentioned that I use bonk~ in some patches as well. I've had better luck with it though when there's not too many different simultaneous frequencies going on. That's what got me filtering the input first- but I will take another look at it, maybe i just didn't use it's full possibilities.
-p
On Sep 1, 2006, at 3:41 PM, Kyle Klipowicz wrote:
Instead of filters, try using bonk~, which you can train tto differentiate various attacks and label them accordingly in its output.
~Kyle
On 9/1/06, Paris Treantafeles paris@parisgraphics.com wrote:
Hello List,
I have some Gem patches that react to incoming sound in various
ways.
Up until now, I would take the adc~ input and use 3 vcf~ 's
(one for
bass, mid and high) and with the right settings of Hz and Q, I was getting fairly good envelopes. Looking at this again and wanting to improve it, I have been experimenting with iemlib filters: http://pd.iem.at/iemlib/abstract.html
I was wondering if anyone can recommend which of these would be
best
for trying to isolate, say a bass drum or a hi hat from an incoming mix?
I read the Bessel filters are often used in cross overs but
from just
experimenting a bit it seems some of the others (e.g. vcf_lp2~)
are
giving me a better response.
Thanks, p
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
--
http://theradioproject.com http://perhapsidid.blogspot.com
(((())))(()()((((((((()())))()(((((((())()()())()))) (())))))(()))))))))))))(((((((((((()()))))))))((()))) ))(((((((((((())))())))))))))))))))__________ _____())))))(((((((((((((()))))))))))_______ ((((((())))))))))))((((((((000)))oOOOOOO
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
--
http://theradioproject.com http://perhapsidid.blogspot.com
(((())))(()()((((((((()())))()(((((((())()()())()))) (())))))(()))))))))))))(((((((((((()()))))))))((()))) ))(((((((((((())))())))))))))))))))__________ _____())))))(((((((((((((()))))))))))_______ ((((((())))))))))))((((((((000)))oOOOOOO
Hey all,
Could [bonk~] be used to recognize human voice? (rather than music)
Seems to me there is not really any effective way to extract voice from music... Has anyone tried this?
.b.
Kyle Klipowicz wrote:
Paris~
I really would recommend the "learn" method, it creates a fft-based profile (I think using a neural net, correct me someone if I'm mistaken) of the sounds through a training routine of a suggested 10 occurrences.
I have not played with this feature too much myself, but imagine that if it works even halfway decent, it could do much to better your applications. Please keep the list updated on your success/failure with this feature.
~Kyle
On 9/1/06, Paris Treantafeles paris@parisgraphics.com wrote:
Hi Kyle,
Thanks - I should have mentioned that I use bonk~ in some patches as well. I've had better luck with it though when there's not too many different simultaneous frequencies going on. That's what got me filtering the input first- but I will take another look at it, maybe i just didn't use it's full possibilities.
-p
On Sep 1, 2006, at 3:41 PM, Kyle Klipowicz wrote:
Instead of filters, try using bonk~, which you can train tto differentiate various attacks and label them accordingly in its output.
~Kyle
On 9/1/06, Paris Treantafeles paris@parisgraphics.com wrote:
Hello List,
I have some Gem patches that react to incoming sound in various ways. Up until now, I would take the adc~ input and use 3 vcf~ 's (one for bass, mid and high) and with the right settings of Hz and Q, I was getting fairly good envelopes. Looking at this again and wanting to improve it, I have been experimenting with iemlib filters: http://pd.iem.at/iemlib/abstract.html
I was wondering if anyone can recommend which of these would be best for trying to isolate, say a bass drum or a hi hat from an incoming mix?
I read the Bessel filters are often used in cross overs but from just experimenting a bit it seems some of the others (e.g. vcf_lp2~) are giving me a better response.
Thanks, p
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
--
http://theradioproject.com http://perhapsidid.blogspot.com
(((())))(()()((((((((()())))()(((((((())()()())()))) (())))))(()))))))))))))(((((((((((()()))))))))((()))) ))(((((((((((())))())))))))))))))))__________ _____())))))(((((((((((((()))))))))))_______ ((((((())))))))))))((((((((000)))oOOOOOO
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
I don't think that bonk~ has the capacity to do this. It's a difficult trick that would require some really keen AI and training. This topic was recently on the .microsound list, due to this article:
http://www.techweb.com/wire/security/192300009
Pretty neat, and scary too, depending on who uses the technology and why.
~Kyle
On 9/2/06, B. Bogart ben@ekran.org wrote:
Hey all,
Could [bonk~] be used to recognize human voice? (rather than music)
Seems to me there is not really any effective way to extract voice from music... Has anyone tried this?
.b.
Kyle Klipowicz wrote:
Paris~
I really would recommend the "learn" method, it creates a fft-based profile (I think using a neural net, correct me someone if I'm mistaken) of the sounds through a training routine of a suggested 10 occurrences.
I have not played with this feature too much myself, but imagine that if it works even halfway decent, it could do much to better your applications. Please keep the list updated on your success/failure with this feature.
~Kyle
On 9/1/06, Paris Treantafeles paris@parisgraphics.com wrote:
Hi Kyle,
Thanks - I should have mentioned that I use bonk~ in some patches as well. I've had better luck with it though when there's not too many different simultaneous frequencies going on. That's what got me filtering the input first- but I will take another look at it, maybe i just didn't use it's full possibilities.
-p
On Sep 1, 2006, at 3:41 PM, Kyle Klipowicz wrote:
Instead of filters, try using bonk~, which you can train tto differentiate various attacks and label them accordingly in its output.
~Kyle
On 9/1/06, Paris Treantafeles paris@parisgraphics.com wrote:
Hello List,
I have some Gem patches that react to incoming sound in various ways. Up until now, I would take the adc~ input and use 3 vcf~ 's (one for bass, mid and high) and with the right settings of Hz and Q, I was getting fairly good envelopes. Looking at this again and wanting to improve it, I have been experimenting with iemlib filters: http://pd.iem.at/iemlib/abstract.html
I was wondering if anyone can recommend which of these would be best for trying to isolate, say a bass drum or a hi hat from an incoming mix?
I read the Bessel filters are often used in cross overs but from just experimenting a bit it seems some of the others (e.g. vcf_lp2~) are giving me a better response.
Thanks, p
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
--
http://theradioproject.com http://perhapsidid.blogspot.com
(((())))(()()((((((((()())))()(((((((())()()())()))) (())))))(()))))))))))))(((((((((((()()))))))))((()))) ))(((((((((((())))())))))))))))))))__________ _____())))))(((((((((((((()))))))))))_______ ((((((())))))))))))((((((((000)))oOOOOOO
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
Kyle Klipowicz a écrit :
I don't think that bonk~ has the capacity to do this. It's a difficult trick that would require some really keen AI and training. This topic was recently on the .microsound list, due to this article:
http://www.techweb.com/wire/security/192300009
Pretty neat, and scary too, depending on who uses the technology and why.
~Kyle
On 9/2/06, B. Bogart ben@ekran.org wrote:
Hey all,
Could [bonk~] be used to recognize human voice? (rather than music)
http://www.tldp.org/HOWTO/Speech-Recognition-HOWTO/
p4.vert.ukl.yahoo.com uncompressed Sat Sep 2 18:27:00 GMT 2006
___________________________________________________________________________
Découvrez un nouveau moyen de poser toutes vos questions quelque soit le sujet !
Yahoo! Questions/Réponses pour partager vos connaissances, vos opinions et vos expériences.
http://fr.answers.yahoo.com
Hi Ben,
I have tried but not had much success. The other way around (cutting the voice out of a mix) is much easier. If anyone wants to share some bonk~ examples that would be cool :-)
-p
On Sep 2, 2006, at 1:03 PM, B. Bogart wrote:
Hey all,
Could [bonk~] be used to recognize human voice? (rather than music)
Seems to me there is not really any effective way to extract voice
from music... Has anyone tried this?.b.
Kyle Klipowicz wrote:
Paris~
I really would recommend the "learn" method, it creates a fft-based profile (I think using a neural net, correct me someone if I'm mistaken) of the sounds through a training routine of a suggested 10 occurrences.
I have not played with this feature too much myself, but imagine that if it works even halfway decent, it could do much to better your applications. Please keep the list updated on your success/failure with this feature.
~Kyle
On 9/1/06, Paris Treantafeles paris@parisgraphics.com wrote:
Hi Kyle,
Thanks - I should have mentioned that I use bonk~ in some patches as well. I've had better luck with it though when there's not too many different simultaneous frequencies going on. That's what got me filtering the input first- but I will take
another look at it, maybe i just didn't use it's full possibilities.-p
On Sep 1, 2006, at 3:41 PM, Kyle Klipowicz wrote:
Instead of filters, try using bonk~, which you can train tto differentiate various attacks and label them accordingly in its output.
~Kyle
On 9/1/06, Paris Treantafeles paris@parisgraphics.com wrote:
Hello List,
I have some Gem patches that react to incoming sound in various
ways. Up until now, I would take the adc~ input and use 3 vcf~ 's
(one for bass, mid and high) and with the right settings of Hz and Q, I was getting fairly good envelopes. Looking at this again and wanting to improve it, I have been experimenting with iemlib filters: http://pd.iem.at/iemlib/abstract.htmlI was wondering if anyone can recommend which of these would be
best for trying to isolate, say a bass drum or a hi hat from an incoming mix?I read the Bessel filters are often used in cross overs but
from just experimenting a bit it seems some of the others (e.g. vcf_lp2~)
are giving me a better response.Thanks, p
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
--
http://theradioproject.com http://perhapsidid.blogspot.com
(((())))(()()((((((((()())))()(((((((())()()())()))) (())))))(()))))))))))))(((((((((((()()))))))))((()))) ))(((((((((((())))())))))))))))))))__________ _____())))))(((((((((((((()))))))))))_______ ((((((())))))))))))((((((((000)))oOOOOOO
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
Hi Paris. If I'm lucky I'll see you at piksel this year.
Note I'm not trying to recognize the speech (as in what it is saying symbolically) but differentiating the speech from the background. Actually I want to extract human voice, not music, from radio signals...
.b.
Paris Treantafeles wrote:
Hi Ben,
I have tried but not had much success. The other way around (cutting the voice out of a mix) is much easier. If anyone wants to share some bonk~ examples that would be cool :-)
-p
On Sep 2, 2006, at 1:03 PM, B. Bogart wrote:
Hey all,
Could [bonk~] be used to recognize human voice? (rather than music)
Seems to me there is not really any effective way to extract voice from music... Has anyone tried this?
.b.
Kyle Klipowicz wrote:
Paris~
I really would recommend the "learn" method, it creates a fft-based profile (I think using a neural net, correct me someone if I'm mistaken) of the sounds through a training routine of a suggested 10 occurrences.
I have not played with this feature too much myself, but imagine that if it works even halfway decent, it could do much to better your applications. Please keep the list updated on your success/failure with this feature.
~Kyle
On 9/1/06, Paris Treantafeles paris@parisgraphics.com wrote:
Hi Kyle,
Thanks - I should have mentioned that I use bonk~ in some patches as well. I've had better luck with it though when there's not too many different simultaneous frequencies going on. That's what got me filtering the input first- but I will take another look at it, maybe i just didn't use it's full possibilities.
-p
On Sep 1, 2006, at 3:41 PM, Kyle Klipowicz wrote:
Instead of filters, try using bonk~, which you can train tto differentiate various attacks and label them accordingly in its output.
~Kyle
On 9/1/06, Paris Treantafeles paris@parisgraphics.com wrote:
Hello List,
I have some Gem patches that react to incoming sound in various ways. Up until now, I would take the adc~ input and use 3 vcf~ 's (one for bass, mid and high) and with the right settings of Hz and Q, I was getting fairly good envelopes. Looking at this again and wanting to improve it, I have been experimenting with iemlib filters: http://pd.iem.at/iemlib/abstract.html
I was wondering if anyone can recommend which of these would be best for trying to isolate, say a bass drum or a hi hat from an incoming mix?
I read the Bessel filters are often used in cross overs but from just experimenting a bit it seems some of the others (e.g. vcf_lp2~) are giving me a better response.
Thanks, p
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
--
http://theradioproject.com http://perhapsidid.blogspot.com
(((())))(()()((((((((()())))()(((((((())()()())()))) (())))))(()))))))))))))(((((((((((()()))))))))((()))) ))(((((((((((())))())))))))))))))))__________ _____())))))(((((((((((((()))))))))))_______ ((((((())))))))))))((((((((000)))oOOOOOO
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
-
B. Bogart -
Kyle Klipowicz -
Paris Treantafeles -
Patco