Hello all,
An user reported a bug in the forum about some problematic behaviours of certain filters in pd-extended 0.43.4, particularly [hip~] and [bp~]. I found this bug to be also present in pd vanilla 0.46.5. Basically, these filters do not respond to frequencies higher than approximately 12 kHz, in contrast to [vcf~] which responds to any frequency. This can be very clearly observed in the case of [bp~] using a narrow band frequency sweep.
Here is a video showing the problem: http://youtu.be/A7mPhiHcmhc Here is the topic in our forum where this discussion started: http://forum.pdpatchrepo.info/topic/8967/highpass-hip-not-working-as-expecte... Here is that user's bug report concerning pd-extended: https://sourceforge.net/p/pure-data/bugs/1187/ Finally, here is the patch I am using to test this (I am using [spectrogram~] (which in its turn needs [metrum] and [multiplex~], all of which are included in this compacted file): https://www.dropbox.com/s/fcwciwe8klwh5xz/filter-test.zip?dl=0
NOTE: the user that reported the bug argues that pd vanilla 0.45.4 does not produce any problems, but I firmly believe this to be a mistake of him. I tested these filters in both 0.45.4 and 0.46.5 and both produce the same behaviour. I tried to contact him to ask for further tests (he was not using my patch above for his own tests), but so far I got no reply. Also, other users confirmed this bug to be present in both vanilla and extended.
Please let me know if I can be of any help with this.
Best, Gilberto Agostinho
it's how the filters were designed, they're not perfect throughout the spectrum
you'll need to use/build other filters for that purpose of working with that higher portion of the spectrum (in which I can hardly still hear, by the way)/
cheers
2015-01-29 20:28 GMT-02:00 Gilberto Agostinho < gilbertohasnofb@googlemail.com>:
Hello all,
An user reported a bug in the forum about some problematic behaviours of certain filters in pd-extended 0.43.4, particularly [hip~] and [bp~]. I found this bug to be also present in pd vanilla 0.46.5. Basically, these filters do not respond to frequencies higher than approximately 12 kHz, in contrast to [vcf~] which responds to any frequency. This can be very clearly observed in the case of [bp~] using a narrow band frequency sweep.
Here is a video showing the problem: http://youtu.be/A7mPhiHcmhc Here is the topic in our forum where this discussion started: http://forum.pdpatchrepo.info/topic/8967/highpass-hip-not- working-as-expected/18 Here is that user's bug report concerning pd-extended: https://sourceforge.net/p/pure-data/bugs/1187/ Finally, here is the patch I am using to test this (I am using [spectrogram~] (which in its turn needs [metrum] and [multiplex~], all of which are included in this compacted file): https://www.dropbox.com/s/ fcwciwe8klwh5xz/filter-test.zip?dl=0
NOTE: the user that reported the bug argues that pd vanilla 0.45.4 does not produce any problems, but I firmly believe this to be a mistake of him. I tested these filters in both 0.45.4 and 0.46.5 and both produce the same behaviour. I tried to contact him to ask for further tests (he was not using my patch above for his own tests), but so far I got no reply. Also, other users confirmed this bug to be present in both vanilla and extended.
Please let me know if I can be of any help with this.
Best, Gilberto Agostinho
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
a few of patches of mine that describe the behaviour of vanilla and some extended filters
2015-01-29 23:39 GMT-02:00 Alexandre Torres Porres porres@gmail.com:
it's how the filters were designed, they're not perfect throughout the spectrum
you'll need to use/build other filters for that purpose of working with that higher portion of the spectrum (in which I can hardly still hear, by the way)/
cheers
2015-01-29 20:28 GMT-02:00 Gilberto Agostinho < gilbertohasnofb@googlemail.com>:
Hello all,
An user reported a bug in the forum about some problematic behaviours of certain filters in pd-extended 0.43.4, particularly [hip~] and [bp~]. I found this bug to be also present in pd vanilla 0.46.5. Basically, these filters do not respond to frequencies higher than approximately 12 kHz, in contrast to [vcf~] which responds to any frequency. This can be very clearly observed in the case of [bp~] using a narrow band frequency sweep.
Here is a video showing the problem: http://youtu.be/A7mPhiHcmhc Here is the topic in our forum where this discussion started: http://forum.pdpatchrepo.info/topic/8967/highpass-hip-not- working-as-expected/18 Here is that user's bug report concerning pd-extended: https://sourceforge.net/p/pure-data/bugs/1187/ Finally, here is the patch I am using to test this (I am using [spectrogram~] (which in its turn needs [metrum] and [multiplex~], all of which are included in this compacted file): https://www.dropbox.com/s/ fcwciwe8klwh5xz/filter-test.zip?dl=0
NOTE: the user that reported the bug argues that pd vanilla 0.45.4 does not produce any problems, but I firmly believe this to be a mistake of him. I tested these filters in both 0.45.4 and 0.46.5 and both produce the same behaviour. I tried to contact him to ask for further tests (he was not using my patch above for his own tests), but so far I got no reply. Also, other users confirmed this bug to be present in both vanilla and extended.
Please let me know if I can be of any help with this.
Best, Gilberto Agostinho
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/ listinfo/pd-list
Hi Alexandre, thanks for your reply.
On 30/01/15 02:39, Alexandre Torres Porres wrote:
it's how the filters were designed, they're not perfect throughout the spectrum
Probably it would be better then to warn users about this limit of ~12 kHz in the help-files then.
you'll need to use/build other filters for that purpose of working with that higher portion of the spectrum (in which I can hardly still hear, by the way)/
Hmm, I can clearly hear the difference between feeding a high frequency to [vcf~] and to [bp~] using a large enough Q (~100 for instance). With that Q, try sweeping from 12kHz until 14kHz in both filters, and you'll hear that [vcf~] behaves as expected, while [bp~] does not change the frequency but only affects the amplitude of the output for some reason.
a few of patches of mine that describe the behaviour of vanilla and some extended filters
Thanks a lot for the patches, but I am actually aware of [vcf~] and [biquad~]. This message of mine was intended to be a bug report rather than a cry for help as I have no particular problem with filters needing a solution right now. But in any case I will check your patches, as I know (and enjoy very much) yours /Computer Music With Examples In Pd/.
Take care, Gilberto Agostinho
This message of mine was intended to be a bug report
I see, but my point is that it's not much of a "bug", so if you want to see how these filters work more deeply, you may check my stuff.
You can see how I made versions of the filters as [fexpr~] or [biquad~] and the raw filters. I made [vcf~] out of [cpole~], for example. And you can check the plot of magnitude/phase.
I know (and enjoy very much) yours
*> Computer Music With Examples In Pd*.
Thanks. That got old though... I translated from portuguese what I had for a workshop at the last pd convention in 2011, but it's mostly FFT. My current portuguese version is way bigger, up to 250 examples in the next release I'm planning for next month. These patches I'm sharing is from my newest releases.
Probably it would be better then to warn users about this limit of ~12 kHz in the help-files then.
Yeah, it'd be good to point out a bit of its characteristics. I learned a lot by checking on my own.
I could realize how [bp~] is quite similar to the resonant lowpass from [lores~] for example. Its just a matter of setting up the gain. If you set [bp~]'s Q to 1, you'll also see how it behaves much like a lowpass filter like [lop~]. The [vcf~] filter has similar characteristics.
cheers
2015-01-30 0:19 GMT-02:00 Gilberto Agostinho <gilbertohasnofb@googlemail.com
:
Hi Alexandre, thanks for your reply.
On 30/01/15 02:39, Alexandre Torres Porres wrote:
it's how the filters were designed, they're not perfect throughout the spectrum
Probably it would be better then to warn users about this limit of ~12 kHz in the help-files then.
you'll need to use/build other filters for that purpose of working with that higher portion of the spectrum (in which I can hardly still hear, by the way)/
Hmm, I can clearly hear the difference between feeding a high frequency to [vcf~] and to [bp~] using a large enough Q (~100 for instance). With that Q, try sweeping from 12kHz until 14kHz in both filters, and you'll hear that [vcf~] behaves as expected, while [bp~] does not change the frequency but only affects the amplitude of the output for some reason.
a few of patches of mine that describe the behaviour of vanilla and some extended filters
Thanks a lot for the patches, but I am actually aware of [vcf~] and [biquad~]. This message of mine was intended to be a bug report rather than a cry for help as I have no particular problem with filters needing a solution right now. But in any case I will check your patches, as I know (and enjoy very much) yours *Computer Music With Examples In Pd*.
Take care, Gilberto Agostinho
Attached patch illustrates how [iemlib/hp1~] is calculated. This gives a perfect hipass, [hip~] could (or should) be done the same way I guess.
Katja
On Fri, Jan 30, 2015 at 6:21 AM, Alexandre Torres Porres porres@gmail.com wrote:
This message of mine was intended to be a bug report
I see, but my point is that it's not much of a "bug", so if you want to see how these filters work more deeply, you may check my stuff.
You can see how I made versions of the filters as [fexpr~] or [biquad~] and the raw filters. I made [vcf~] out of [cpole~], for example. And you can check the plot of magnitude/phase.
I know (and enjoy very much) yours Computer Music With Examples In Pd.
Thanks. That got old though... I translated from portuguese what I had for a workshop at the last pd convention in 2011, but it's mostly FFT. My current portuguese version is way bigger, up to 250 examples in the next release I'm planning for next month. These patches I'm sharing is from my newest releases.
Probably it would be better then to warn users about this limit of ~12 kHz in the help-files then.
Yeah, it'd be good to point out a bit of its characteristics. I learned a lot by checking on my own.
I could realize how [bp~] is quite similar to the resonant lowpass from [lores~] for example. Its just a matter of setting up the gain. If you set [bp~]'s Q to 1, you'll also see how it behaves much like a lowpass filter like [lop~]. The [vcf~] filter has similar characteristics.
cheers
2015-01-30 0:19 GMT-02:00 Gilberto Agostinho gilbertohasnofb@googlemail.com:
Hi Alexandre, thanks for your reply.
On 30/01/15 02:39, Alexandre Torres Porres wrote:
it's how the filters were designed, they're not perfect throughout the spectrum
Probably it would be better then to warn users about this limit of ~12 kHz in the help-files then.
you'll need to use/build other filters for that purpose of working with that higher portion of the spectrum (in which I can hardly still hear, by the way)/
Hmm, I can clearly hear the difference between feeding a high frequency to [vcf~] and to [bp~] using a large enough Q (~100 for instance). With that Q, try sweeping from 12kHz until 14kHz in both filters, and you'll hear that [vcf~] behaves as expected, while [bp~] does not change the frequency but only affects the amplitude of the output for some reason.
a few of patches of mine that describe the behaviour of vanilla and some extended filters
Thanks a lot for the patches, but I am actually aware of [vcf~] and [biquad~]. This message of mine was intended to be a bug report rather than a cry for help as I have no particular problem with filters needing a solution right now. But in any case I will check your patches, as I know (and enjoy very much) yours Computer Music With Examples In Pd.
Take care, Gilberto Agostinho
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
that seems quite nicer than [hip~], cool.
I rewrote it to make it more concise with expr.
cheers
2015-01-30 19:29 GMT-02:00 katja katjavetter@gmail.com:
Attached patch illustrates how [iemlib/hp1~] is calculated. This gives a perfect hipass, [hip~] could (or should) be done the same way I guess.
Katja
On Fri, Jan 30, 2015 at 6:21 AM, Alexandre Torres Porres porres@gmail.com wrote:
This message of mine was intended to be a bug report
I see, but my point is that it's not much of a "bug", so if you want to
see
how these filters work more deeply, you may check my stuff.
You can see how I made versions of the filters as [fexpr~] or [biquad~]
and
the raw filters. I made [vcf~] out of [cpole~], for example. And you can check the plot of magnitude/phase.
I know (and enjoy very much) yours Computer Music With Examples In Pd.
Thanks. That got old though... I translated from portuguese what I had
for a
workshop at the last pd convention in 2011, but it's mostly FFT. My
current
portuguese version is way bigger, up to 250 examples in the next release
I'm
planning for next month. These patches I'm sharing is from my newest releases.
Probably it would be better then to warn users about this limit of ~12 kHz in the help-files then.
Yeah, it'd be good to point out a bit of its characteristics. I learned a lot by checking on my own.
I could realize how [bp~] is quite similar to the resonant lowpass from [lores~] for example. Its just a matter of setting up the gain. If you
set
[bp~]'s Q to 1, you'll also see how it behaves much like a lowpass filter like [lop~]. The [vcf~] filter has similar characteristics.
cheers
2015-01-30 0:19 GMT-02:00 Gilberto Agostinho gilbertohasnofb@googlemail.com:
Hi Alexandre, thanks for your reply.
On 30/01/15 02:39, Alexandre Torres Porres wrote:
it's how the filters were designed, they're not perfect throughout the spectrum
Probably it would be better then to warn users about this limit of ~12
kHz
in the help-files then.
you'll need to use/build other filters for that purpose of working with that higher portion of the spectrum (in which I can hardly still hear,
by
the way)/
Hmm, I can clearly hear the difference between feeding a high frequency
to
[vcf~] and to [bp~] using a large enough Q (~100 for instance). With
that Q,
try sweeping from 12kHz until 14kHz in both filters, and you'll hear
that
[vcf~] behaves as expected, while [bp~] does not change the frequency
but
only affects the amplitude of the output for some reason.
a few of patches of mine that describe the behaviour of vanilla and some extended filters
Thanks a lot for the patches, but I am actually aware of [vcf~] and [biquad~]. This message of mine was intended to be a bug report rather
than
a cry for help as I have no particular problem with filters needing a solution right now. But in any case I will check your patches, as I know (and enjoy very much) yours Computer Music With Examples In Pd.
Take care, Gilberto Agostinho
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
and here's hip with raw filters
2015-01-30 20:09 GMT-02:00 Alexandre Torres Porres porres@gmail.com:
that seems quite nicer than [hip~], cool.
I rewrote it to make it more concise with expr.
cheers
2015-01-30 19:29 GMT-02:00 katja katjavetter@gmail.com:
Attached patch illustrates how [iemlib/hp1~] is calculated. This gives
a perfect hipass, [hip~] could (or should) be done the same way I guess.
Katja
On Fri, Jan 30, 2015 at 6:21 AM, Alexandre Torres Porres porres@gmail.com wrote:
This message of mine was intended to be a bug report
I see, but my point is that it's not much of a "bug", so if you want to
see
how these filters work more deeply, you may check my stuff.
You can see how I made versions of the filters as [fexpr~] or [biquad~]
and
the raw filters. I made [vcf~] out of [cpole~], for example. And you can check the plot of magnitude/phase.
I know (and enjoy very much) yours Computer Music With Examples In Pd.
Thanks. That got old though... I translated from portuguese what I had
for a
workshop at the last pd convention in 2011, but it's mostly FFT. My
current
portuguese version is way bigger, up to 250 examples in the next
release I'm
planning for next month. These patches I'm sharing is from my newest releases.
Probably it would be better then to warn users about this limit of ~12 kHz in the help-files then.
Yeah, it'd be good to point out a bit of its characteristics. I learned
a
lot by checking on my own.
I could realize how [bp~] is quite similar to the resonant lowpass from [lores~] for example. Its just a matter of setting up the gain. If you
set
[bp~]'s Q to 1, you'll also see how it behaves much like a lowpass
filter
like [lop~]. The [vcf~] filter has similar characteristics.
cheers
2015-01-30 0:19 GMT-02:00 Gilberto Agostinho gilbertohasnofb@googlemail.com:
Hi Alexandre, thanks for your reply.
On 30/01/15 02:39, Alexandre Torres Porres wrote:
it's how the filters were designed, they're not perfect throughout the spectrum
Probably it would be better then to warn users about this limit of ~12
kHz
in the help-files then.
you'll need to use/build other filters for that purpose of working with that higher portion of the spectrum (in which I can hardly still hear,
by
the way)/
Hmm, I can clearly hear the difference between feeding a high
frequency to
[vcf~] and to [bp~] using a large enough Q (~100 for instance). With
that Q,
try sweeping from 12kHz until 14kHz in both filters, and you'll hear
that
[vcf~] behaves as expected, while [bp~] does not change the frequency
but
only affects the amplitude of the output for some reason.
a few of patches of mine that describe the behaviour of vanilla and
some
extended filters
Thanks a lot for the patches, but I am actually aware of [vcf~] and [biquad~]. This message of mine was intended to be a bug report rather
than
a cry for help as I have no particular problem with filters needing a solution right now. But in any case I will check your patches, as I
know
(and enjoy very much) yours Computer Music With Examples In Pd.
Take care, Gilberto Agostinho
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
-
Alexandre Torres Porres -
Gilberto Agostinho -
katja