I have a strange problem while working with pvsbufread. The example  
below crashes Csound ("Csound tidy up: Segmentation fault") (I'm on  
Csound 5.10 on OSX Intel 10.5.6).
When I change the ksmps to 10, 11, ...15 it works.
I checked all the ksmps from 1 to 16: 4, 8, and 16 are crashing.
So i thought there might be a problem with power-of-two values.
But:
When I change p3 to 4 or 5 (instead of 3), it works also with ksmps=16.
So I begin believing in ghosts ...
Thanks for any help keeping my mind in the republic of  
rationality ... :)
	joachim




-odac -iadc


sr = 44100
ksmps = 16
nchnls = 1
0dbfs = 1
instr 1; resording in a pvs-buffer
ain			inch			1
fsigin		pvsanal		ain, 1024, 256, 1024, 0
gifftbuf, knull	pvsbuffer		fsigin, 10
endin
instr 2; plays the pvs-buffer
ktime		linseg		0, p3, p3
fsig  			pvsbufread   	ktime, gifftbuf
aout			pvsynth		fsig
			outch		1, aout
endin


i 1 0 3
i 2 1 3

I'll have to investigate that one later. Thanks for
reporting it. I think it might have to do with ktime.

One thing I'll suggest is to use the time reference
given by pvsbuffer and alter it to the timestretching
or compressing. If that goes with your design,
that is.

Regards
Victor

----- Original Message ----- 
From: "joachim heintz" 
To: 
Sent: Saturday, May 23, 2009 1:27 PM
Subject: [Csnd] pvsbufread, ksmps and ghosts over p3


>I have a strange problem while working with pvsbufread. The example  below 
>crashes Csound ("Csound tidy up: Segmentation fault") (I'm on  Csound 5.10 
>on OSX Intel 10.5.6).
> When I change the ksmps to 10, 11, ...15 it works.
> I checked all the ksmps from 1 to 16: 4, 8, and 16 are crashing.
> So i thought there might be a problem with power-of-two values.
> But:
> When I change p3 to 4 or 5 (instead of 3), it works also with ksmps=16.
> So I begin believing in ghosts ...
> Thanks for any help keeping my mind in the republic of  rationality ... :)
> joachim
>
>
> 
> 
> -odac -iadc
> 
> 
> sr = 44100
> ksmps = 16
> nchnls = 1
> 0dbfs = 1
> instr 1; resording in a pvs-buffer
> ain inch 1
> fsigin pvsanal ain, 1024, 256, 1024, 0
> gifftbuf, knull pvsbuffer fsigin, 10
> endin
> instr 2; plays the pvs-buffer
> ktime linseg 0, p3, p3
> fsig  pvsbufread   ktime, gifftbuf
> aout pvsynth fsig
> outch 1, aout
> endin
> 
> 
> i 1 0 3
> i 2 1 3
> 
> 
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
> csound" 

I think I fixed this. In CVS now.
----- Original Message ----- 
From: "victor" 
To: 
Sent: Saturday, May 23, 2009 1:35 PM
Subject: [Csnd] Re: pvsbufread, ksmps and ghosts over p3


> I'll have to investigate that one later. Thanks for
> reporting it. I think it might have to do with ktime.
>
> One thing I'll suggest is to use the time reference
> given by pvsbuffer and alter it to the timestretching
> or compressing. If that goes with your design,
> that is.
>
> Regards
> Victor
>
> ----- Original Message ----- 
> From: "joachim heintz" 
> To: 
> Sent: Saturday, May 23, 2009 1:27 PM
> Subject: [Csnd] pvsbufread, ksmps and ghosts over p3
>
>
>>I have a strange problem while working with pvsbufread. The example  below 
>>crashes Csound ("Csound tidy up: Segmentation fault") (I'm on  Csound 5.10 
>>on OSX Intel 10.5.6).
>> When I change the ksmps to 10, 11, ...15 it works.
>> I checked all the ksmps from 1 to 16: 4, 8, and 16 are crashing.
>> So i thought there might be a problem with power-of-two values.
>> But:
>> When I change p3 to 4 or 5 (instead of 3), it works also with ksmps=16.
>> So I begin believing in ghosts ...
>> Thanks for any help keeping my mind in the republic of  rationality ... 
>> :)
>> joachim
>>
>>
>> 
>> 
>> -odac -iadc
>> 
>> 
>> sr = 44100
>> ksmps = 16
>> nchnls = 1
>> 0dbfs = 1
>> instr 1; resording in a pvs-buffer
>> ain inch 1
>> fsigin pvsanal ain, 1024, 256, 1024, 0
>> gifftbuf, knull pvsbuffer fsigin, 10
>> endin
>> instr 2; plays the pvs-buffer
>> ktime linseg 0, p3, p3
>> fsig  pvsbufread   ktime, gifftbuf
>> aout pvsynth fsig
>> outch 1, aout
>> endin
>> 
>> 
>> i 1 0 3
>> i 2 1 3
>> 
>> 
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
>> csound"
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
> csound" 

Thanks. I'm working with it for a live project. My solution at the  
moment is to set ksmps=2. This is stable (though no very efficient).
One more notice: When I use non powof2 ksmps (for instance 10, or 3),  
the quality of "playback" in my example is very bad.
Can this be related to the issue, or is there something I have to do  
to smooth the resynthesizing?
Best -

	joachim


Am 23.05.2009 um 14:35 schrieb victor:

> I'll have to investigate that one later. Thanks for
> reporting it. I think it might have to do with ktime.
>
> One thing I'll suggest is to use the time reference
> given by pvsbuffer and alter it to the timestretching
> or compressing. If that goes with your design,
> that is.
>
> Regards
> Victor
>
> ----- Original Message ----- From: "joachim heintz"  >
> To: 
> Sent: Saturday, May 23, 2009 1:27 PM
> Subject: [Csnd] pvsbufread, ksmps and ghosts over p3
>
>
>> I have a strange problem while working with pvsbufread. The  
>> example  below crashes Csound ("Csound tidy up: Segmentation  
>> fault") (I'm on  Csound 5.10 on OSX Intel 10.5.6).
>> When I change the ksmps to 10, 11, ...15 it works.
>> I checked all the ksmps from 1 to 16: 4, 8, and 16 are crashing.
>> So i thought there might be a problem with power-of-two values.
>> But:
>> When I change p3 to 4 or 5 (instead of 3), it works also with  
>> ksmps=16.
>> So I begin believing in ghosts ...
>> Thanks for any help keeping my mind in the republic of   
>> rationality ... :)
>> joachim
>>
>>
>> 
>> 
>> -odac -iadc
>> 
>> 
>> sr = 44100
>> ksmps = 16
>> nchnls = 1
>> 0dbfs = 1
>> instr 1; resording in a pvs-buffer
>> ain inch 1
>> fsigin pvsanal ain, 1024, 256, 1024, 0
>> gifftbuf, knull pvsbuffer fsigin, 10
>> endin
>> instr 2; plays the pvs-buffer
>> ktime linseg 0, p3, p3
>> fsig  pvsbufread   ktime, gifftbuf
>> aout pvsynth fsig
>> outch 1, aout
>> endin
>> 
>> 
>> i 1 0 3
>> i 2 1 3
>> 
>> 
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>> "unsubscribe csound"
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
> "unsubscribe csound"
>

I noticed you are using a Hamming window, I would suggest
hanning, which will give you better result. Not sure about the
pow of two, but I noticed kmsps = 32 does not crash it or
if you make the duration 3.01 secs. Why do you need such
a low ksmps? 64 or 128 work OK for me.

Victor
----- Original Message ----- 
From: "joachim heintz" 
To: 
Sent: Saturday, May 23, 2009 2:11 PM
Subject: [Csnd] Re: Re: pvsbufread, ksmps and ghosts over p3


> Thanks. I'm working with it for a live project. My solution at the  moment 
> is to set ksmps=2. This is stable (though no very efficient).
> One more notice: When I use non powof2 ksmps (for instance 10, or 3),  the 
> quality of "playback" in my example is very bad.
> Can this be related to the issue, or is there something I have to do  to 
> smooth the resynthesizing?
> Best -
>
> joachim
>
>
> Am 23.05.2009 um 14:35 schrieb victor:
>
>> I'll have to investigate that one later. Thanks for
>> reporting it. I think it might have to do with ktime.
>>
>> One thing I'll suggest is to use the time reference
>> given by pvsbuffer and alter it to the timestretching
>> or compressing. If that goes with your design,
>> that is.
>>
>> Regards
>> Victor
>>
>> ----- Original Message ----- From: "joachim heintz" > >
>> To: 
>> Sent: Saturday, May 23, 2009 1:27 PM
>> Subject: [Csnd] pvsbufread, ksmps and ghosts over p3
>>
>>
>>> I have a strange problem while working with pvsbufread. The  example 
>>> below crashes Csound ("Csound tidy up: Segmentation  fault") (I'm on 
>>> Csound 5.10 on OSX Intel 10.5.6).
>>> When I change the ksmps to 10, 11, ...15 it works.
>>> I checked all the ksmps from 1 to 16: 4, 8, and 16 are crashing.
>>> So i thought there might be a problem with power-of-two values.
>>> But:
>>> When I change p3 to 4 or 5 (instead of 3), it works also with  ksmps=16.
>>> So I begin believing in ghosts ...
>>> Thanks for any help keeping my mind in the republic of   rationality ... 
>>> :)
>>> joachim
>>>
>>>
>>> 
>>> 
>>> -odac -iadc
>>> 
>>> 
>>> sr = 44100
>>> ksmps = 16
>>> nchnls = 1
>>> 0dbfs = 1
>>> instr 1; resording in a pvs-buffer
>>> ain inch 1
>>> fsigin pvsanal ain, 1024, 256, 1024, 0
>>> gifftbuf, knull pvsbuffer fsigin, 10
>>> endin
>>> instr 2; plays the pvs-buffer
>>> ktime linseg 0, p3, p3
>>> fsig  pvsbufread   ktime, gifftbuf
>>> aout pvsynth fsig
>>> outch 1, aout
>>> endin
>>> 
>>> 
>>> i 1 0 3
>>> i 2 1 3
>>> 
>>> 
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>> "unsubscribe csound"
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  "unsubscribe 
>> csound"
>>
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
> csound" 

I think the audio quality might be to do with the
fact that for full-duplex -B and -b have to be
multiples of ksmps.

Victor
----- Original Message ----- 
From: "joachim heintz" 
To: 
Sent: Saturday, May 23, 2009 2:11 PM
Subject: [Csnd] Re: Re: pvsbufread, ksmps and ghosts over p3


> Thanks. I'm working with it for a live project. My solution at the  moment 
> is to set ksmps=2. This is stable (though no very efficient).
> One more notice: When I use non powof2 ksmps (for instance 10, or 3),  the 
> quality of "playback" in my example is very bad.
> Can this be related to the issue, or is there something I have to do  to 
> smooth the resynthesizing?
> Best -
>
> joachim
>
>
> Am 23.05.2009 um 14:35 schrieb victor:
>
>> I'll have to investigate that one later. Thanks for
>> reporting it. I think it might have to do with ktime.
>>
>> One thing I'll suggest is to use the time reference
>> given by pvsbuffer and alter it to the timestretching
>> or compressing. If that goes with your design,
>> that is.
>>
>> Regards
>> Victor
>>
>> ----- Original Message ----- From: "joachim heintz" > >
>> To: 
>> Sent: Saturday, May 23, 2009 1:27 PM
>> Subject: [Csnd] pvsbufread, ksmps and ghosts over p3
>>
>>
>>> I have a strange problem while working with pvsbufread. The  example 
>>> below crashes Csound ("Csound tidy up: Segmentation  fault") (I'm on 
>>> Csound 5.10 on OSX Intel 10.5.6).
>>> When I change the ksmps to 10, 11, ...15 it works.
>>> I checked all the ksmps from 1 to 16: 4, 8, and 16 are crashing.
>>> So i thought there might be a problem with power-of-two values.
>>> But:
>>> When I change p3 to 4 or 5 (instead of 3), it works also with  ksmps=16.
>>> So I begin believing in ghosts ...
>>> Thanks for any help keeping my mind in the republic of   rationality ... 
>>> :)
>>> joachim
>>>
>>>
>>> 
>>> 
>>> -odac -iadc
>>> 
>>> 
>>> sr = 44100
>>> ksmps = 16
>>> nchnls = 1
>>> 0dbfs = 1
>>> instr 1; resording in a pvs-buffer
>>> ain inch 1
>>> fsigin pvsanal ain, 1024, 256, 1024, 0
>>> gifftbuf, knull pvsbuffer fsigin, 10
>>> endin
>>> instr 2; plays the pvs-buffer
>>> ktime linseg 0, p3, p3
>>> fsig  pvsbufread   ktime, gifftbuf
>>> aout pvsynth fsig
>>> outch 1, aout
>>> endin
>>> 
>>> 
>>> i 1 0 3
>>> i 2 1 3
>>> 
>>> 
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>> "unsubscribe csound"
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  "unsubscribe 
>> csound"
>>
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
> csound" 

Thanks for the tip with the von-Hann window. As to the ksmps, in my  
experience the time resolution for envelopes with small durations is  
much too bad when I use a large ksmps. Very audible results then  
(could be avoided by using a-rate envelopes, but usually I'm on the k- 
level there).
Isn't this time resolution the reason for the worse audio quality when  
using a nonpowof2 ksmps? I would expect the windows (which are powof2)  
going out of synchronisation at the overlaps, aren't they?

	joachim



Am 23.05.2009 um 17:48 schrieb victor:

> I noticed you are using a Hamming window, I would suggest
> hanning, which will give you better result. Not sure about the
> pow of two, but I noticed kmsps = 32 does not crash it or
> if you make the duration 3.01 secs. Why do you need such
> a low ksmps? 64 or 128 work OK for me.
>
> Victor
> ----- Original Message ----- From: "joachim heintz"  >
> To: 
> Sent: Saturday, May 23, 2009 2:11 PM
> Subject: [Csnd] Re: Re: pvsbufread, ksmps and ghosts over p3
>
>
>> Thanks. I'm working with it for a live project. My solution at the   
>> moment is to set ksmps=2. This is stable (though no very efficient).
>> One more notice: When I use non powof2 ksmps (for instance 10, or  
>> 3),  the quality of "playback" in my example is very bad.
>> Can this be related to the issue, or is there something I have to  
>> do  to smooth the resynthesizing?
>> Best -
>>
>> joachim
>>
>>
>> Am 23.05.2009 um 14:35 schrieb victor:
>>
>>> I'll have to investigate that one later. Thanks for
>>> reporting it. I think it might have to do with ktime.
>>>
>>> One thing I'll suggest is to use the time reference
>>> given by pvsbuffer and alter it to the timestretching
>>> or compressing. If that goes with your design,
>>> that is.
>>>
>>> Regards
>>> Victor
>>>
>>> ----- Original Message ----- From: "joachim heintz" >> >
>>> To: 
>>> Sent: Saturday, May 23, 2009 1:27 PM
>>> Subject: [Csnd] pvsbufread, ksmps and ghosts over p3
>>>
>>>
>>>> I have a strange problem while working with pvsbufread. The   
>>>> example below crashes Csound ("Csound tidy up: Segmentation   
>>>> fault") (I'm on Csound 5.10 on OSX Intel 10.5.6).
>>>> When I change the ksmps to 10, 11, ...15 it works.
>>>> I checked all the ksmps from 1 to 16: 4, 8, and 16 are crashing.
>>>> So i thought there might be a problem with power-of-two values.
>>>> But:
>>>> When I change p3 to 4 or 5 (instead of 3), it works also with   
>>>> ksmps=16.
>>>> So I begin believing in ghosts ...
>>>> Thanks for any help keeping my mind in the republic of    
>>>> rationality ... :)
>>>> joachim
>>>>
>>>>
>>>> 
>>>> 
>>>> -odac -iadc
>>>> 
>>>> 
>>>> sr = 44100
>>>> ksmps = 16
>>>> nchnls = 1
>>>> 0dbfs = 1
>>>> instr 1; resording in a pvs-buffer
>>>> ain inch 1
>>>> fsigin pvsanal ain, 1024, 256, 1024, 0
>>>> gifftbuf, knull pvsbuffer fsigin, 10
>>>> endin
>>>> instr 2; plays the pvs-buffer
>>>> ktime linseg 0, p3, p3
>>>> fsig  pvsbufread   ktime, gifftbuf
>>>> aout pvsynth fsig
>>>> outch 1, aout
>>>> endin
>>>> 
>>>> 
>>>> i 1 0 3
>>>> i 2 1 3
>>>> 
>>>> 
>>>>
>>>>
>>>> Send bugs reports to this list.
>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>>> "unsubscribe csound"
>>>
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body   
>>> "unsubscribe csound"
>>>
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>> "unsubscribe csound"
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
> "unsubscribe csound"
>

No, I think the non-pow-of-two quality problem is not related
to pvs in particular, but with the IO buffering in full-duplex audio.
In this case buffers should be multiples of ksmps. If the buffer
is pow-of-two, then making ksmps pow-of-two achieves this.

Victor

----- Original Message ----- 
From: "joachim heintz" 
To: 
Sent: Saturday, May 23, 2009 10:17 PM
Subject: [Csnd] Re: Re: Re: Re: pvsbufread, ksmps and ghosts over p3


> Thanks for the tip with the von-Hann window. As to the ksmps, in my 
> experience the time resolution for envelopes with small durations is  much 
> too bad when I use a large ksmps. Very audible results then  (could be 
> avoided by using a-rate envelopes, but usually I'm on the k- level there).
> Isn't this time resolution the reason for the worse audio quality when 
> using a nonpowof2 ksmps? I would expect the windows (which are powof2) 
> going out of synchronisation at the overlaps, aren't they?
>
> joachim
>
>
>
> Am 23.05.2009 um 17:48 schrieb victor:
>
>> I noticed you are using a Hamming window, I would suggest
>> hanning, which will give you better result. Not sure about the
>> pow of two, but I noticed kmsps = 32 does not crash it or
>> if you make the duration 3.01 secs. Why do you need such
>> a low ksmps? 64 or 128 work OK for me.
>>
>> Victor
>> ----- Original Message ----- From: "joachim heintz" > >
>> To: 
>> Sent: Saturday, May 23, 2009 2:11 PM
>> Subject: [Csnd] Re: Re: pvsbufread, ksmps and ghosts over p3
>>
>>
>>> Thanks. I'm working with it for a live project. My solution at the 
>>> moment is to set ksmps=2. This is stable (though no very efficient).
>>> One more notice: When I use non powof2 ksmps (for instance 10, or  3), 
>>> the quality of "playback" in my example is very bad.
>>> Can this be related to the issue, or is there something I have to  do 
>>> to smooth the resynthesizing?
>>> Best -
>>>
>>> joachim
>>>
>>>
>>> Am 23.05.2009 um 14:35 schrieb victor:
>>>
>>>> I'll have to investigate that one later. Thanks for
>>>> reporting it. I think it might have to do with ktime.
>>>>
>>>> One thing I'll suggest is to use the time reference
>>>> given by pvsbuffer and alter it to the timestretching
>>>> or compressing. If that goes with your design,
>>>> that is.
>>>>
>>>> Regards
>>>> Victor
>>>>
>>>> ----- Original Message ----- From: "joachim heintz" 
>>>> >>> >
>>>> To: 
>>>> Sent: Saturday, May 23, 2009 1:27 PM
>>>> Subject: [Csnd] pvsbufread, ksmps and ghosts over p3
>>>>
>>>>
>>>>> I have a strange problem while working with pvsbufread. The   example 
>>>>> below crashes Csound ("Csound tidy up: Segmentation   fault") (I'm on 
>>>>> Csound 5.10 on OSX Intel 10.5.6).
>>>>> When I change the ksmps to 10, 11, ...15 it works.
>>>>> I checked all the ksmps from 1 to 16: 4, 8, and 16 are crashing.
>>>>> So i thought there might be a problem with power-of-two values.
>>>>> But:
>>>>> When I change p3 to 4 or 5 (instead of 3), it works also with 
>>>>> ksmps=16.
>>>>> So I begin believing in ghosts ...
>>>>> Thanks for any help keeping my mind in the republic of    rationality 
>>>>> ... :)
>>>>> joachim
>>>>>
>>>>>
>>>>> 
>>>>> 
>>>>> -odac -iadc
>>>>> 
>>>>> 
>>>>> sr = 44100
>>>>> ksmps = 16
>>>>> nchnls = 1
>>>>> 0dbfs = 1
>>>>> instr 1; resording in a pvs-buffer
>>>>> ain inch 1
>>>>> fsigin pvsanal ain, 1024, 256, 1024, 0
>>>>> gifftbuf, knull pvsbuffer fsigin, 10
>>>>> endin
>>>>> instr 2; plays the pvs-buffer
>>>>> ktime linseg 0, p3, p3
>>>>> fsig  pvsbufread   ktime, gifftbuf
>>>>> aout pvsynth fsig
>>>>> outch 1, aout
>>>>> endin
>>>>> 
>>>>> 
>>>>> i 1 0 3
>>>>> i 2 1 3
>>>>> 
>>>>> 
>>>>>
>>>>>
>>>>> Send bugs reports to this list.
>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>>>> "unsubscribe csound"
>>>>
>>>>
>>>>
>>>> Send bugs reports to this list.
>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>>> "unsubscribe csound"
>>>>
>>>
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>> "unsubscribe csound"
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  "unsubscribe 
>> csound"
>>
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
> csound" 

This is an important point for me. Could you perhaps explain which  
consequences there are if one uses full-duplex audio or not? Until now  
I thought it's just a question of efficiency, but it seems to be more?!
Thanks -

	joachim


Am 24.05.2009 um 00:22 schrieb victor:

> No, I think the non-pow-of-two quality problem is not related
> to pvs in particular, but with the IO buffering in full-duplex audio.
> In this case buffers should be multiples of ksmps. If the buffer
> is pow-of-two, then making ksmps pow-of-two achieves this.
>
> Victor
>
> ----- Original Message ----- From: "joachim heintz"  >
> To: 
> Sent: Saturday, May 23, 2009 10:17 PM
> Subject: [Csnd] Re: Re: Re: Re: pvsbufread, ksmps and ghosts over p3
>
>
>> Thanks for the tip with the von-Hann window. As to the ksmps, in my  
>> experience the time resolution for envelopes with small durations  
>> is  much too bad when I use a large ksmps. Very audible results  
>> then  (could be avoided by using a-rate envelopes, but usually I'm  
>> on the k- level there).
>> Isn't this time resolution the reason for the worse audio quality  
>> when using a nonpowof2 ksmps? I would expect the windows (which are  
>> powof2) going out of synchronisation at the overlaps, aren't they?
>>
>> joachim
>>
>>
>>
>> Am 23.05.2009 um 17:48 schrieb victor:
>>
>>> I noticed you are using a Hamming window, I would suggest
>>> hanning, which will give you better result. Not sure about the
>>> pow of two, but I noticed kmsps = 32 does not crash it or
>>> if you make the duration 3.01 secs. Why do you need such
>>> a low ksmps? 64 or 128 work OK for me.
>>>
>>> Victor
>>> ----- Original Message ----- From: "joachim heintz" >> >
>>> To: 
>>> Sent: Saturday, May 23, 2009 2:11 PM
>>> Subject: [Csnd] Re: Re: pvsbufread, ksmps and ghosts over p3
>>>
>>>
>>>> Thanks. I'm working with it for a live project. My solution at  
>>>> the moment is to set ksmps=2. This is stable (though no very  
>>>> efficient).
>>>> One more notice: When I use non powof2 ksmps (for instance 10,  
>>>> or  3), the quality of "playback" in my example is very bad.
>>>> Can this be related to the issue, or is there something I have  
>>>> to  do to smooth the resynthesizing?
>>>> Best -
>>>>
>>>> joachim
>>>>
>>>>
>>>> Am 23.05.2009 um 14:35 schrieb victor:
>>>>
>>>>> I'll have to investigate that one later. Thanks for
>>>>> reporting it. I think it might have to do with ktime.
>>>>>
>>>>> One thing I'll suggest is to use the time reference
>>>>> given by pvsbuffer and alter it to the timestretching
>>>>> or compressing. If that goes with your design,
>>>>> that is.
>>>>>
>>>>> Regards
>>>>> Victor
>>>>>
>>>>> ----- Original Message ----- From: "joachim heintz" >>>> >
>>>>> To: 
>>>>> Sent: Saturday, May 23, 2009 1:27 PM
>>>>> Subject: [Csnd] pvsbufread, ksmps and ghosts over p3
>>>>>
>>>>>
>>>>>> I have a strange problem while working with pvsbufread. The    
>>>>>> example below crashes Csound ("Csound tidy up: Segmentation    
>>>>>> fault") (I'm on Csound 5.10 on OSX Intel 10.5.6).
>>>>>> When I change the ksmps to 10, 11, ...15 it works.
>>>>>> I checked all the ksmps from 1 to 16: 4, 8, and 16 are crashing.
>>>>>> So i thought there might be a problem with power-of-two values.
>>>>>> But:
>>>>>> When I change p3 to 4 or 5 (instead of 3), it works also with  
>>>>>> ksmps=16.
>>>>>> So I begin believing in ghosts ...
>>>>>> Thanks for any help keeping my mind in the republic of     
>>>>>> rationality ... :)
>>>>>> joachim
>>>>>>
>>>>>>
>>>>>> 
>>>>>> 
>>>>>> -odac -iadc
>>>>>> 
>>>>>> 
>>>>>> sr = 44100
>>>>>> ksmps = 16
>>>>>> nchnls = 1
>>>>>> 0dbfs = 1
>>>>>> instr 1; resording in a pvs-buffer
>>>>>> ain inch 1
>>>>>> fsigin pvsanal ain, 1024, 256, 1024, 0
>>>>>> gifftbuf, knull pvsbuffer fsigin, 10
>>>>>> endin
>>>>>> instr 2; plays the pvs-buffer
>>>>>> ktime linseg 0, p3, p3
>>>>>> fsig  pvsbufread   ktime, gifftbuf
>>>>>> aout pvsynth fsig
>>>>>> outch 1, aout
>>>>>> endin
>>>>>> 
>>>>>> 
>>>>>> i 1 0 3
>>>>>> i 2 1 3
>>>>>> 
>>>>>> 
>>>>>>
>>>>>>
>>>>>> Send bugs reports to this list.
>>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>>>>> "unsubscribe csound"
>>>>>
>>>>>
>>>>>
>>>>> Send bugs reports to this list.
>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>>>> "unsubscribe csound"
>>>>>
>>>>
>>>>
>>>>
>>>> Send bugs reports to this list.
>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>>> "unsubscribe csound"
>>>
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body   
>>> "unsubscribe csound"
>>>
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>> "unsubscribe csound"
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
> "unsubscribe csound"
>

The restriction is when you use -idac -odac, then
if the buffers are not multiples of the ksmps, there are
synchronisation issues and you get dropped frames.
So the rule of thumb is: use ksmps = pow of two,
since -b and -B often are pow of two too.

With just -odac there is no such restriction

Victor

----- Original Message ----- 
From: "joachim heintz" 
To: 
Sent: Sunday, May 24, 2009 4:58 PM
Subject: [Csnd] Re: Re: Re: Re: Re: Re: pvsbufread, ksmps and ghosts over p3


> This is an important point for me. Could you perhaps explain which 
> consequences there are if one uses full-duplex audio or not? Until now  I 
> thought it's just a question of efficiency, but it seems to be more?!
> Thanks -
>
> joachim
>
>
> Am 24.05.2009 um 00:22 schrieb victor:
>
>> No, I think the non-pow-of-two quality problem is not related
>> to pvs in particular, but with the IO buffering in full-duplex audio.
>> In this case buffers should be multiples of ksmps. If the buffer
>> is pow-of-two, then making ksmps pow-of-two achieves this.
>>
>> Victor
>>
>> ----- Original Message ----- From: "joachim heintz" > >
>> To: 
>> Sent: Saturday, May 23, 2009 10:17 PM
>> Subject: [Csnd] Re: Re: Re: Re: pvsbufread, ksmps and ghosts over p3
>>
>>
>>> Thanks for the tip with the von-Hann window. As to the ksmps, in my 
>>> experience the time resolution for envelopes with small durations  is 
>>> much too bad when I use a large ksmps. Very audible results  then 
>>> (could be avoided by using a-rate envelopes, but usually I'm  on the k- 
>>> level there).
>>> Isn't this time resolution the reason for the worse audio quality  when 
>>> using a nonpowof2 ksmps? I would expect the windows (which are  powof2) 
>>> going out of synchronisation at the overlaps, aren't they?
>>>
>>> joachim
>>>
>>>
>>>
>>> Am 23.05.2009 um 17:48 schrieb victor:
>>>
>>>> I noticed you are using a Hamming window, I would suggest
>>>> hanning, which will give you better result. Not sure about the
>>>> pow of two, but I noticed kmsps = 32 does not crash it or
>>>> if you make the duration 3.01 secs. Why do you need such
>>>> a low ksmps? 64 or 128 work OK for me.
>>>>
>>>> Victor
>>>> ----- Original Message ----- From: "joachim heintz" 
>>>> >>> >
>>>> To: 
>>>> Sent: Saturday, May 23, 2009 2:11 PM
>>>> Subject: [Csnd] Re: Re: pvsbufread, ksmps and ghosts over p3
>>>>
>>>>
>>>>> Thanks. I'm working with it for a live project. My solution at  the 
>>>>> moment is to set ksmps=2. This is stable (though no very  efficient).
>>>>> One more notice: When I use non powof2 ksmps (for instance 10,  or 
>>>>> 3), the quality of "playback" in my example is very bad.
>>>>> Can this be related to the issue, or is there something I have  to  do 
>>>>> to smooth the resynthesizing?
>>>>> Best -
>>>>>
>>>>> joachim
>>>>>
>>>>>
>>>>> Am 23.05.2009 um 14:35 schrieb victor:
>>>>>
>>>>>> I'll have to investigate that one later. Thanks for
>>>>>> reporting it. I think it might have to do with ktime.
>>>>>>
>>>>>> One thing I'll suggest is to use the time reference
>>>>>> given by pvsbuffer and alter it to the timestretching
>>>>>> or compressing. If that goes with your design,
>>>>>> that is.
>>>>>>
>>>>>> Regards
>>>>>> Victor
>>>>>>
>>>>>> ----- Original Message ----- From: "joachim heintz" 
>>>>>> >>>>> >
>>>>>> To: 
>>>>>> Sent: Saturday, May 23, 2009 1:27 PM
>>>>>> Subject: [Csnd] pvsbufread, ksmps and ghosts over p3
>>>>>>
>>>>>>
>>>>>>> I have a strange problem while working with pvsbufread. The 
>>>>>>> example below crashes Csound ("Csound tidy up: Segmentation 
>>>>>>> fault") (I'm on Csound 5.10 on OSX Intel 10.5.6).
>>>>>>> When I change the ksmps to 10, 11, ...15 it works.
>>>>>>> I checked all the ksmps from 1 to 16: 4, 8, and 16 are crashing.
>>>>>>> So i thought there might be a problem with power-of-two values.
>>>>>>> But:
>>>>>>> When I change p3 to 4 or 5 (instead of 3), it works also with 
>>>>>>> ksmps=16.
>>>>>>> So I begin believing in ghosts ...
>>>>>>> Thanks for any help keeping my mind in the republic of 
>>>>>>> rationality ... :)
>>>>>>> joachim
>>>>>>>
>>>>>>>
>>>>>>> 
>>>>>>> 
>>>>>>> -odac -iadc
>>>>>>> 
>>>>>>> 
>>>>>>> sr = 44100
>>>>>>> ksmps = 16
>>>>>>> nchnls = 1
>>>>>>> 0dbfs = 1
>>>>>>> instr 1; resording in a pvs-buffer
>>>>>>> ain inch 1
>>>>>>> fsigin pvsanal ain, 1024, 256, 1024, 0
>>>>>>> gifftbuf, knull pvsbuffer fsigin, 10
>>>>>>> endin
>>>>>>> instr 2; plays the pvs-buffer
>>>>>>> ktime linseg 0, p3, p3
>>>>>>> fsig  pvsbufread   ktime, gifftbuf
>>>>>>> aout pvsynth fsig
>>>>>>> outch 1, aout
>>>>>>> endin
>>>>>>> 
>>>>>>> 
>>>>>>> i 1 0 3
>>>>>>> i 2 1 3
>>>>>>> 
>>>>>>> 
>>>>>>>
>>>>>>>
>>>>>>> Send bugs reports to this list.
>>>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>>>>>> "unsubscribe csound"
>>>>>>
>>>>>>
>>>>>>
>>>>>> Send bugs reports to this list.
>>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>>>>> "unsubscribe csound"
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>> Send bugs reports to this list.
>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>>>> "unsubscribe csound"
>>>>
>>>>
>>>>
>>>> Send bugs reports to this list.
>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>>> "unsubscribe csound"
>>>>
>>>
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>> "unsubscribe csound"
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  "unsubscribe 
>> csound"
>>
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
> csound" 

Thanks for the explanation, Victor. So this is really important for  
live situations. Best -

	joachim



Am 24.05.2009 um 21:22 schrieb victor:

> The restriction is when you use -idac -odac, then
> if the buffers are not multiples of the ksmps, there are
> synchronisation issues and you get dropped frames.
> So the rule of thumb is: use ksmps = pow of two,
> since -b and -B often are pow of two too.
>
> With just -odac there is no such restriction
>
> Victor
>
> ----- Original Message ----- From: "joachim heintz"  >
> To: 
> Sent: Sunday, May 24, 2009 4:58 PM
> Subject: [Csnd] Re: Re: Re: Re: Re: Re: pvsbufread, ksmps and ghosts  
> over p3
>
>
>> This is an important point for me. Could you perhaps explain which  
>> consequences there are if one uses full-duplex audio or not? Until  
>> now  I thought it's just a question of efficiency, but it seems to  
>> be more?!
>> Thanks -
>>
>> joachim
>>
>>
>> Am 24.05.2009 um 00:22 schrieb victor:
>>
>>> No, I think the non-pow-of-two quality problem is not related
>>> to pvs in particular, but with the IO buffering in full-duplex  
>>> audio.
>>> In this case buffers should be multiples of ksmps. If the buffer
>>> is pow-of-two, then making ksmps pow-of-two achieves this.
>>>
>>> Victor
>>>
>>> ----- Original Message ----- From: "joachim heintz" >> >
>>> To: 
>>> Sent: Saturday, May 23, 2009 10:17 PM
>>> Subject: [Csnd] Re: Re: Re: Re: pvsbufread, ksmps and ghosts over p3
>>>
>>>
>>>> Thanks for the tip with the von-Hann window. As to the ksmps, in  
>>>> my experience the time resolution for envelopes with small  
>>>> durations  is much too bad when I use a large ksmps. Very audible  
>>>> results  then (could be avoided by using a-rate envelopes, but  
>>>> usually I'm  on the k- level there).
>>>> Isn't this time resolution the reason for the worse audio  
>>>> quality  when using a nonpowof2 ksmps? I would expect the windows  
>>>> (which are  powof2) going out of synchronisation at the overlaps,  
>>>> aren't they?
>>>>
>>>> joachim
>>>>
>>>>
>>>>
>>>> Am 23.05.2009 um 17:48 schrieb victor:
>>>>
>>>>> I noticed you are using a Hamming window, I would suggest
>>>>> hanning, which will give you better result. Not sure about the
>>>>> pow of two, but I noticed kmsps = 32 does not crash it or
>>>>> if you make the duration 3.01 secs. Why do you need such
>>>>> a low ksmps? 64 or 128 work OK for me.
>>>>>
>>>>> Victor
>>>>> ----- Original Message ----- From: "joachim heintz" >>>> >
>>>>> To: 
>>>>> Sent: Saturday, May 23, 2009 2:11 PM
>>>>> Subject: [Csnd] Re: Re: pvsbufread, ksmps and ghosts over p3
>>>>>
>>>>>
>>>>>> Thanks. I'm working with it for a live project. My solution at   
>>>>>> the moment is to set ksmps=2. This is stable (though no very   
>>>>>> efficient).
>>>>>> One more notice: When I use non powof2 ksmps (for instance 10,   
>>>>>> or 3), the quality of "playback" in my example is very bad.
>>>>>> Can this be related to the issue, or is there something I have   
>>>>>> to  do to smooth the resynthesizing?
>>>>>> Best -
>>>>>>
>>>>>> joachim
>>>>>>
>>>>>>
>>>>>> Am 23.05.2009 um 14:35 schrieb victor:
>>>>>>
>>>>>>> I'll have to investigate that one later. Thanks for
>>>>>>> reporting it. I think it might have to do with ktime.
>>>>>>>
>>>>>>> One thing I'll suggest is to use the time reference
>>>>>>> given by pvsbuffer and alter it to the timestretching
>>>>>>> or compressing. If that goes with your design,
>>>>>>> that is.
>>>>>>>
>>>>>>> Regards
>>>>>>> Victor
>>>>>>>
>>>>>>> ----- Original Message ----- From: "joachim heintz" >>>>>> >
>>>>>>> To: 
>>>>>>> Sent: Saturday, May 23, 2009 1:27 PM
>>>>>>> Subject: [Csnd] pvsbufread, ksmps and ghosts over p3
>>>>>>>
>>>>>>>
>>>>>>>> I have a strange problem while working with pvsbufread. The  
>>>>>>>> example below crashes Csound ("Csound tidy up: Segmentation  
>>>>>>>> fault") (I'm on Csound 5.10 on OSX Intel 10.5.6).
>>>>>>>> When I change the ksmps to 10, 11, ...15 it works.
>>>>>>>> I checked all the ksmps from 1 to 16: 4, 8, and 16 are  
>>>>>>>> crashing.
>>>>>>>> So i thought there might be a problem with power-of-two values.
>>>>>>>> But:
>>>>>>>> When I change p3 to 4 or 5 (instead of 3), it works also with  
>>>>>>>> ksmps=16.
>>>>>>>> So I begin believing in ghosts ...
>>>>>>>> Thanks for any help keeping my mind in the republic of  
>>>>>>>> rationality ... :)
>>>>>>>> joachim
>>>>>>>>
>>>>>>>>
>>>>>>>> 
>>>>>>>> 
>>>>>>>> -odac -iadc
>>>>>>>> 
>>>>>>>> 
>>>>>>>> sr = 44100
>>>>>>>> ksmps = 16
>>>>>>>> nchnls = 1
>>>>>>>> 0dbfs = 1
>>>>>>>> instr 1; resording in a pvs-buffer
>>>>>>>> ain inch 1
>>>>>>>> fsigin pvsanal ain, 1024, 256, 1024, 0
>>>>>>>> gifftbuf, knull pvsbuffer fsigin, 10
>>>>>>>> endin
>>>>>>>> instr 2; plays the pvs-buffer
>>>>>>>> ktime linseg 0, p3, p3
>>>>>>>> fsig  pvsbufread   ktime, gifftbuf
>>>>>>>> aout pvsynth fsig
>>>>>>>> outch 1, aout
>>>>>>>> endin
>>>>>>>> 
>>>>>>>> 
>>>>>>>> i 1 0 3
>>>>>>>> i 2 1 3
>>>>>>>> 
>>>>>>>> 
>>>>>>>>
>>>>>>>>
>>>>>>>> Send bugs reports to this list.
>>>>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>>>>>>> "unsubscribe csound"
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Send bugs reports to this list.
>>>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>>>>>> "unsubscribe csound"
>>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> Send bugs reports to this list.
>>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>>>>> "unsubscribe csound"
>>>>>
>>>>>
>>>>>
>>>>> Send bugs reports to this list.
>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>>>> "unsubscribe csound"
>>>>>
>>>>
>>>>
>>>>
>>>> Send bugs reports to this list.
>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>>> "unsubscribe csound"
>>>
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body   
>>> "unsubscribe csound"
>>>
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>> "unsubscribe csound"
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
> "unsubscribe csound"
>

Hi,

Would there be synchroniation issues in every case? I would think
there would be synchronization issues if you're using two separate
cards with different clocks...

Cheers,
Andrés

On Sun, May 24, 2009 at 2:22 PM, victor  wrote:
> The restriction is when you use -idac -odac, then
> if the buffers are not multiples of the ksmps, there are
> synchronisation issues and you get dropped frames.
> So the rule of thumb is: use ksmps = pow of two,
> since -b and -B often are pow of two too.
>
> With just -odac there is no such restriction
>
> Victor
>
> ----- Original Message ----- From: "joachim heintz" 
> To: 
> Sent: Sunday, May 24, 2009 4:58 PM
> Subject: [Csnd] Re: Re: Re: Re: Re: Re: pvsbufread, ksmps and ghosts over p3
>
>
>> This is an important point for me. Could you perhaps explain which
>> consequences there are if one uses full-duplex audio or not? Until now  I
>> thought it's just a question of efficiency, but it seems to be more?!
>> Thanks -
>>
>> joachim
>>
>>
>> Am 24.05.2009 um 00:22 schrieb victor:
>>
>>> No, I think the non-pow-of-two quality problem is not related
>>> to pvs in particular, but with the IO buffering in full-duplex audio.
>>> In this case buffers should be multiples of ksmps. If the buffer
>>> is pow-of-two, then making ksmps pow-of-two achieves this.
>>>
>>> Victor
>>>
>>> ----- Original Message ----- From: "joachim heintz" >> >
>>> To: 
>>> Sent: Saturday, May 23, 2009 10:17 PM
>>> Subject: [Csnd] Re: Re: Re: Re: pvsbufread, ksmps and ghosts over p3
>>>
>>>
>>>> Thanks for the tip with the von-Hann window. As to the ksmps, in my
>>>> experience the time resolution for envelopes with small durations  is much
>>>> too bad when I use a large ksmps. Very audible results  then (could be
>>>> avoided by using a-rate envelopes, but usually I'm  on the k- level there).
>>>> Isn't this time resolution the reason for the worse audio quality  when
>>>> using a nonpowof2 ksmps? I would expect the windows (which are  powof2)
>>>> going out of synchronisation at the overlaps, aren't they?
>>>>
>>>> joachim
>>>>
>>>>
>>>>
>>>> Am 23.05.2009 um 17:48 schrieb victor:
>>>>
>>>>> I noticed you are using a Hamming window, I would suggest
>>>>> hanning, which will give you better result. Not sure about the
>>>>> pow of two, but I noticed kmsps = 32 does not crash it or
>>>>> if you make the duration 3.01 secs. Why do you need such
>>>>> a low ksmps? 64 or 128 work OK for me.
>>>>>
>>>>> Victor
>>>>> ----- Original Message ----- From: "joachim heintz"
>>>>> >>>> >
>>>>> To: 
>>>>> Sent: Saturday, May 23, 2009 2:11 PM
>>>>> Subject: [Csnd] Re: Re: pvsbufread, ksmps and ghosts over p3
>>>>>
>>>>>
>>>>>> Thanks. I'm working with it for a live project. My solution at  the
>>>>>> moment is to set ksmps=2. This is stable (though no very  efficient).
>>>>>> One more notice: When I use non powof2 ksmps (for instance 10,  or 3),
>>>>>> the quality of "playback" in my example is very bad.
>>>>>> Can this be related to the issue, or is there something I have  to  do
>>>>>> to smooth the resynthesizing?
>>>>>> Best -
>>>>>>
>>>>>> joachim
>>>>>>
>>>>>>
>>>>>> Am 23.05.2009 um 14:35 schrieb victor:
>>>>>>
>>>>>>> I'll have to investigate that one later. Thanks for
>>>>>>> reporting it. I think it might have to do with ktime.
>>>>>>>
>>>>>>> One thing I'll suggest is to use the time reference
>>>>>>> given by pvsbuffer and alter it to the timestretching
>>>>>>> or compressing. If that goes with your design,
>>>>>>> that is.
>>>>>>>
>>>>>>> Regards
>>>>>>> Victor
>>>>>>>
>>>>>>> ----- Original Message ----- From: "joachim heintz"
>>>>>>> >>>>>> >
>>>>>>> To: 
>>>>>>> Sent: Saturday, May 23, 2009 1:27 PM
>>>>>>> Subject: [Csnd] pvsbufread, ksmps and ghosts over p3
>>>>>>>
>>>>>>>
>>>>>>>> I have a strange problem while working with pvsbufread. The example
>>>>>>>> below crashes Csound ("Csound tidy up: Segmentation fault") (I'm on Csound
>>>>>>>> 5.10 on OSX Intel 10.5.6).
>>>>>>>> When I change the ksmps to 10, 11, ...15 it works.
>>>>>>>> I checked all the ksmps from 1 to 16: 4, 8, and 16 are crashing.
>>>>>>>> So i thought there might be a problem with power-of-two values.
>>>>>>>> But:
>>>>>>>> When I change p3 to 4 or 5 (instead of 3), it works also with
>>>>>>>> ksmps=16.
>>>>>>>> So I begin believing in ghosts ...
>>>>>>>> Thanks for any help keeping my mind in the republic of rationality
>>>>>>>> ... :)
>>>>>>>> joachim
>>>>>>>>
>>>>>>>>
>>>>>>>> 
>>>>>>>> 
>>>>>>>> -odac -iadc
>>>>>>>> 
>>>>>>>> 
>>>>>>>> sr = 44100
>>>>>>>> ksmps = 16
>>>>>>>> nchnls = 1
>>>>>>>> 0dbfs = 1
>>>>>>>> instr 1; resording in a pvs-buffer
>>>>>>>> ain inch 1
>>>>>>>> fsigin pvsanal ain, 1024, 256, 1024, 0
>>>>>>>> gifftbuf, knull pvsbuffer fsigin, 10
>>>>>>>> endin
>>>>>>>> instr 2; plays the pvs-buffer
>>>>>>>> ktime linseg 0, p3, p3
>>>>>>>> fsig  pvsbufread   ktime, gifftbuf
>>>>>>>> aout pvsynth fsig
>>>>>>>> outch 1, aout
>>>>>>>> endin
>>>>>>>> 
>>>>>>>> 
>>>>>>>> i 1 0 3
>>>>>>>> i 2 1 3
>>>>>>>> 
>>>>>>>> 
>>>>>>>>
>>>>>>>>
>>>>>>>> Send bugs reports to this list.
>>>>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>>>>>>>> "unsubscribe csound"
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Send bugs reports to this list.
>>>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>>>>>>> "unsubscribe csound"
>>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> Send bugs reports to this list.
>>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>>>>>> "unsubscribe csound"
>>>>>
>>>>>
>>>>>
>>>>> Send bugs reports to this list.
>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>>>>> "unsubscribe csound"
>>>>>
>>>>
>>>>
>>>>
>>>> Send bugs reports to this list.
>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>>>> csound"
>>>
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  "unsubscribe
>>> csound"
>>>
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>> csound"
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
> csound"
>



-- 


Andrés

No it's not that kind of synchronisation. It's to do with how
the IO is implemented and I used synchronisation loosely
here.

Victor
----- Original Message ----- 
From: "Andres Cabrera" 
To: 
Sent: Monday, May 25, 2009 2:03 PM
Subject: [Csnd] Re: Re: Re: Re: Re: Re: Re: Re: pvsbufread, ksmps and ghosts 
over p3


Hi,

Would there be synchroniation issues in every case? I would think
there would be synchronization issues if you're using two separate
cards with different clocks...

Cheers,
Andrés

On Sun, May 24, 2009 at 2:22 PM, victor  wrote:
> The restriction is when you use -idac -odac, then
> if the buffers are not multiples of the ksmps, there are
> synchronisation issues and you get dropped frames.
> So the rule of thumb is: use ksmps = pow of two,
> since -b and -B often are pow of two too.
>
> With just -odac there is no such restriction
>
> Victor
>
> ----- Original Message ----- From: "joachim heintz" 
> To: 
> Sent: Sunday, May 24, 2009 4:58 PM
> Subject: [Csnd] Re: Re: Re: Re: Re: Re: pvsbufread, ksmps and ghosts over 
> p3
>
>
>> This is an important point for me. Could you perhaps explain which
>> consequences there are if one uses full-duplex audio or not? Until now I
>> thought it's just a question of efficiency, but it seems to be more?!
>> Thanks -
>>
>> joachim
>>
>>
>> Am 24.05.2009 um 00:22 schrieb victor:
>>
>>> No, I think the non-pow-of-two quality problem is not related
>>> to pvs in particular, but with the IO buffering in full-duplex audio.
>>> In this case buffers should be multiples of ksmps. If the buffer
>>> is pow-of-two, then making ksmps pow-of-two achieves this.
>>>
>>> Victor
>>>
>>> ----- Original Message ----- From: "joachim heintz" >> >
>>> To: 
>>> Sent: Saturday, May 23, 2009 10:17 PM
>>> Subject: [Csnd] Re: Re: Re: Re: pvsbufread, ksmps and ghosts over p3
>>>
>>>
>>>> Thanks for the tip with the von-Hann window. As to the ksmps, in my
>>>> experience the time resolution for envelopes with small durations is 
>>>> much
>>>> too bad when I use a large ksmps. Very audible results then (could be
>>>> avoided by using a-rate envelopes, but usually I'm on the k- level 
>>>> there).
>>>> Isn't this time resolution the reason for the worse audio quality when
>>>> using a nonpowof2 ksmps? I would expect the windows (which are powof2)
>>>> going out of synchronisation at the overlaps, aren't they?
>>>>
>>>> joachim
>>>>
>>>>
>>>>
>>>> Am 23.05.2009 um 17:48 schrieb victor:
>>>>
>>>>> I noticed you are using a Hamming window, I would suggest
>>>>> hanning, which will give you better result. Not sure about the
>>>>> pow of two, but I noticed kmsps = 32 does not crash it or
>>>>> if you make the duration 3.01 secs. Why do you need such
>>>>> a low ksmps? 64 or 128 work OK for me.
>>>>>
>>>>> Victor
>>>>> ----- Original Message ----- From: "joachim heintz"
>>>>> >>>> >
>>>>> To: 
>>>>> Sent: Saturday, May 23, 2009 2:11 PM
>>>>> Subject: [Csnd] Re: Re: pvsbufread, ksmps and ghosts over p3
>>>>>
>>>>>
>>>>>> Thanks. I'm working with it for a live project. My solution at the
>>>>>> moment is to set ksmps=2. This is stable (though no very efficient).
>>>>>> One more notice: When I use non powof2 ksmps (for instance 10, or 3),
>>>>>> the quality of "playback" in my example is very bad.
>>>>>> Can this be related to the issue, or is there something I have to do
>>>>>> to smooth the resynthesizing?
>>>>>> Best -
>>>>>>
>>>>>> joachim
>>>>>>
>>>>>>
>>>>>> Am 23.05.2009 um 14:35 schrieb victor:
>>>>>>
>>>>>>> I'll have to investigate that one later. Thanks for
>>>>>>> reporting it. I think it might have to do with ktime.
>>>>>>>
>>>>>>> One thing I'll suggest is to use the time reference
>>>>>>> given by pvsbuffer and alter it to the timestretching
>>>>>>> or compressing. If that goes with your design,
>>>>>>> that is.
>>>>>>>
>>>>>>> Regards
>>>>>>> Victor
>>>>>>>
>>>>>>> ----- Original Message ----- From: "joachim heintz"
>>>>>>> >>>>>> >
>>>>>>> To: 
>>>>>>> Sent: Saturday, May 23, 2009 1:27 PM
>>>>>>> Subject: [Csnd] pvsbufread, ksmps and ghosts over p3
>>>>>>>
>>>>>>>
>>>>>>>> I have a strange problem while working with pvsbufread. The example
>>>>>>>> below crashes Csound ("Csound tidy up: Segmentation fault") (I'm on 
>>>>>>>> Csound
>>>>>>>> 5.10 on OSX Intel 10.5.6).
>>>>>>>> When I change the ksmps to 10, 11, ...15 it works.
>>>>>>>> I checked all the ksmps from 1 to 16: 4, 8, and 16 are crashing.
>>>>>>>> So i thought there might be a problem with power-of-two values.
>>>>>>>> But:
>>>>>>>> When I change p3 to 4 or 5 (instead of 3), it works also with
>>>>>>>> ksmps=16.
>>>>>>>> So I begin believing in ghosts ...
>>>>>>>> Thanks for any help keeping my mind in the republic of rationality
>>>>>>>> ... :)
>>>>>>>> joachim
>>>>>>>>
>>>>>>>>
>>>>>>>> 
>>>>>>>> 
>>>>>>>> -odac -iadc
>>>>>>>> 
>>>>>>>> 
>>>>>>>> sr = 44100
>>>>>>>> ksmps = 16
>>>>>>>> nchnls = 1
>>>>>>>> 0dbfs = 1
>>>>>>>> instr 1; resording in a pvs-buffer
>>>>>>>> ain inch 1
>>>>>>>> fsigin pvsanal ain, 1024, 256, 1024, 0
>>>>>>>> gifftbuf, knull pvsbuffer fsigin, 10
>>>>>>>> endin
>>>>>>>> instr 2; plays the pvs-buffer
>>>>>>>> ktime linseg 0, p3, p3
>>>>>>>> fsig pvsbufread ktime, gifftbuf
>>>>>>>> aout pvsynth fsig
>>>>>>>> outch 1, aout
>>>>>>>> endin
>>>>>>>> 
>>>>>>>> 
>>>>>>>> i 1 0 3
>>>>>>>> i 2 1 3
>>>>>>>> 
>>>>>>>> 
>>>>>>>>
>>>>>>>>
>>>>>>>> Send bugs reports to this list.
>>>>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>>>>>>>> "unsubscribe csound"
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Send bugs reports to this list.
>>>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>>>>>>> "unsubscribe csound"
>>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> Send bugs reports to this list.
>>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>>>>>> "unsubscribe csound"
>>>>>
>>>>>
>>>>>
>>>>> Send bugs reports to this list.
>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>>>>> "unsubscribe csound"
>>>>>
>>>>
>>>>
>>>>
>>>> Send bugs reports to this list.
>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body 
>>>> "unsubscribe
>>>> csound"
>>>
>>>
>>>
>>> Send bugs reports to this list.
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>>> csound"
>>>
>>
>>
>>
>> Send bugs reports to this list.
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>> csound"
>
>
>
> Send bugs reports to this list.
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
> csound"
>



-- 


Andrés


Send bugs reports to this list.
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe 
csound"=