I am currently working on a multi channel audio convolver using Csound. I am
creating two ftables and loading an Impulse Response 2048 points long into
them at a sample rate of 48kHz. I am using the first gen function table
commands (iRL ftgen 0, 0, 2048, -1, "IRL.wav", 0, 0, 0) and (iRR ftgen 0, 0,
2048, -1, "IRR.wav", 0, 0, 0 ). 

One problem i currently have is setting the function table number to
negative (-1) does stop scaling but each ftable is scaled differently. One
scalemax value is 0.155 and the other 0.145. This causes a shift in the
stereo image toward the left the (table scaled by 0.155). I was wandering
wether it would be possible to stop this scaling from happening.




--
View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749.html
Sent from the Csound - General mailing list archive at Nabble.com.

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

I have attached a link to my code and also the console output of csound when
running the file. Both the Impulse Responses are Stereo IR's.
https://www.dropbox.com/sh/q5hw3c3aa5jmvk0/AAAOne2ZeLTfV4snhiKV7p7qa?dl=0



--
View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746750.html
Sent from the Csound - General mailing list archive at Nabble.com.

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

Hi Akshay,

I think you posted this question on Stackoverflow and I'll reiterate
one part of Joachim's answer: are you sure the two files for the
Stereo IR don't have two different maxima? Could you post the IR
files?

steven

On Thu, Jan 28, 2016 at 10:53 PM, akshay.rama  wrote:
> I have attached a link to my code and also the console output of csound when
> running the file. Both the Impulse Responses are Stereo IR's.
> https://www.dropbox.com/sh/q5hw3c3aa5jmvk0/AAAOne2ZeLTfV4snhiKV7p7qa?dl=0
>
>
>
> --
> View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746750.html
> Sent from the Csound - General mailing list archive at Nabble.com.
>
> Csound mailing list
> Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> Send bugs reports to
>         https://github.com/csound/csound/issues
> Discussions of bugs and features can be posted here

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

Hi Steven,

Both the files do have a slightly different maxima, but I do not wan't
csound to make this maxima the same since it causes the stereo image to
shift position to either the left or right depending on which is scaled
more. Would it be possible to completely skip the scaling so it uses the two
impulse responses in the same format I provided? 

I was initially using the pconvolve opcode in csound and this worked
perfectly, but I am currently trying to use a function table so I can get
rid of the IR files after csound has loaded it. 



--
View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746752.html
Sent from the Csound - General mailing list archive at Nabble.com.

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

I think the scalemax is reporting the abs max and not a scaling
coefficient (if I interpreted the code it correctly, the message is
reporting what the winascii driver is scaling value by for drawing,
which would be the max values).

I haven't used ftconv before: Are IRL and IRR both stereo wav files?
Also could you post what you used for pconvolve?

On Thu, Jan 28, 2016 at 11:20 PM, akshay.rama  wrote:
> Hi Steven,
>
> Both the files do have a slightly different maxima, but I do not wan't
> csound to make this maxima the same since it causes the stereo image to
> shift position to either the left or right depending on which is scaled
> more. Would it be possible to completely skip the scaling so it uses the two
> impulse responses in the same format I provided?
>
> I was initially using the pconvolve opcode in csound and this worked
> perfectly, but I am currently trying to use a function table so I can get
> rid of the IR files after csound has loaded it.
>
>
>
> --
> View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746752.html
> Sent from the Csound - General mailing list archive at Nabble.com.
>
> Csound mailing list
> Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> Send bugs reports to
>         https://github.com/csound/csound/issues
> Discussions of bugs and features can be posted here

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

Hi Steven,

I have attached the pconvolve code on the dropbox link too.
(https://www.dropbox.com/sh/q5hw3c3aa5jmvk0/AAAOne2ZeLTfV4snhiKV7p7qa?dl=0)

 With the following code, the pconvolve function's audio output is perfectly
centred, but with the ftconv opcode, the stereo image is instantly shifted
slightly to the left. This is why I think the impulse response files are
still being scaled. 

Both IRL and IRR are stereo WAV files. 

The reason i shifted to the ftconv opcode is because I need to be able to
delete the wav impulse response files from the hard disk as soon as csound
initializes and loads the files. With pconv deleting the file causes an
error ( i think this is because it reads through the wav file every time the
filter is activated).

Thanks



--
View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746754.html
Sent from the Csound - General mailing list archive at Nabble.com.

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

Can you provide the IR files so we can investigate?
========================
Dr Victor Lazzarini
Dean of Arts, Celtic Studies and Philosophy,
Maynooth University,
Maynooth, Co Kildare, Ireland
Tel: 00 353 7086936
Fax: 00 353 1 7086952 

> On 29 Jan 2016, at 05:35, akshay.rama  wrote:
> 
> Hi Steven,
> 
> I have attached the pconvolve code on the dropbox link too.
> (https://www.dropbox.com/sh/q5hw3c3aa5jmvk0/AAAOne2ZeLTfV4snhiKV7p7qa?dl=0)
> 
> With the following code, the pconvolve function's audio output is perfectly
> centred, but with the ftconv opcode, the stereo image is instantly shifted
> slightly to the left. This is why I think the impulse response files are
> still being scaled. 
> 
> Both IRL and IRR are stereo WAV files. 
> 
> The reason i shifted to the ftconv opcode is because I need to be able to
> delete the wav impulse response files from the hard disk as soon as csound
> initializes and loads the files. With pconv deleting the file causes an
> error ( i think this is because it reads through the wav file every time the
> filter is activated).
> 
> Thanks
> 
> 
> 
> --
> View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746754.html
> Sent from the Csound - General mailing list archive at Nabble.com.
> 
> Csound mailing list
> Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> Send bugs reports to
>        https://github.com/csound/csound/issues
> Discussions of bugs and features can be posted here

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

yes, scalemax reports the max value at the table.
========================
Dr Victor Lazzarini
Dean of Arts, Celtic Studies and Philosophy,
Maynooth University,
Maynooth, Co Kildare, Ireland
Tel: 00 353 7086936
Fax: 00 353 1 7086952 

> On 29 Jan 2016, at 05:02, Steven Yi  wrote:
> 
> I think the scalemax is reporting the abs max and not a scaling
> coefficient (if I interpreted the code it correctly, the message is
> reporting what the winascii driver is scaling value by for drawing,
> which would be the max values).
> 
> I haven't used ftconv before: Are IRL and IRR both stereo wav files?
> Also could you post what you used for pconvolve?
> 
> On Thu, Jan 28, 2016 at 11:20 PM, akshay.rama  wrote:
>> Hi Steven,
>> 
>> Both the files do have a slightly different maxima, but I do not wan't
>> csound to make this maxima the same since it causes the stereo image to
>> shift position to either the left or right depending on which is scaled
>> more. Would it be possible to completely skip the scaling so it uses the two
>> impulse responses in the same format I provided?
>> 
>> I was initially using the pconvolve opcode in csound and this worked
>> perfectly, but I am currently trying to use a function table so I can get
>> rid of the IR files after csound has loaded it.
>> 
>> 
>> 
>> --
>> View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746752.html
>> Sent from the Csound - General mailing list archive at Nabble.com.
>> 
>> Csound mailing list
>> Csound@listserv.heanet.ie
>> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
>> Send bugs reports to
>>        https://github.com/csound/csound/issues
>> Discussions of bugs and features can be posted here
> 
> Csound mailing list
> Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> Send bugs reports to
>        https://github.com/csound/csound/issues
> Discussions of bugs and features can be posted here

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

another possibility I would try is combining the two stereo IR files into a single 4 channel IR file

you would then use:

gir ftgen 0, 0, 2048, -1, "IR4chwav", 0, 0, 0 ; e.g. chan1,2==irL, chan3,4==irR

then

adummy1, adummy2,ar3 , ar4		ftconv gaR, gir, ipartitionsize, 0, 0, 0 
ar1 , ar2,adummy1,adummy2 		ftconv gaL, gir, ipartitionsize, 0, 0, 0 

this should preserve the relative amplitudes among the channels 

cheers,

Karin

> On 29 Jan 2016, at 06:35, akshay.rama  wrote:
> 
> Hi Steven,
> 
> I have attached the pconvolve code on the dropbox link too.
> (https://www.dropbox.com/sh/q5hw3c3aa5jmvk0/AAAOne2ZeLTfV4snhiKV7p7qa?dl=0)
> 
> With the following code, the pconvolve function's audio output is perfectly
> centred, but with the ftconv opcode, the stereo image is instantly shifted
> slightly to the left. This is why I think the impulse response files are
> still being scaled. 
> 
> Both IRL and IRR are stereo WAV files. 
> 
> The reason i shifted to the ftconv opcode is because I need to be able to
> delete the wav impulse response files from the hard disk as soon as csound
> initializes and loads the files. With pconv deleting the file causes an
> error ( i think this is because it reads through the wav file every time the
> filter is activated).
> 
> Thanks
> 
> 
> 
> --
> View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746754.html
> Sent from the Csound - General mailing list archive at Nabble.com.
> 
> Csound mailing list
> Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> Send bugs reports to
>        https://github.com/csound/csound/issues
> Discussions of bugs and features can be posted here

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

That is a good idea, then you don’t even need to suppress rescaling. In any case, I just 
went and checkrf all the steps with a stereo IR I created, and I could not find any problems,
the outputs of pconvolve and convolve are identical. Rescaling is suppressed as expected.

This is the test. “ir.wav” is any stereo IR, and “input.wav” is any mono input file.

nchnls=2
0dbfs=1

gi1 ftgen 1,0,0,-1,"ir.wav",0,0,1
gi2 ftgen 2,0,0,-1,"ir.wav”,0,0,2

instr 1
a1 diskin “input.wav"
a2 ftconv a1,1,512
a3 ftconv a1,2,512
   outs a2,a3
endin
instr 2
a1 diskin “input.wav"
a2,a3 pconvolve a1,"ir.wav",512
   outs a2,a3
endin

So whatever the problem is, it is not with rescaling, and not with ftconv.

Regards
========================
Dr Victor Lazzarini
Dean of Arts, Celtic Studies and Philosophy,
Maynooth University,
Maynooth, Co Kildare, Ireland
Tel: 00 353 7086936
Fax: 00 353 1 7086952 

> On 29 Jan 2016, at 09:23, Karin Daum  wrote:
> 
> another possibility I would try is combining the two stereo IR files into a single 4 channel IR file
> 
> you would then use:
> 
> gir ftgen 0, 0, 2048, -1, "IR4chwav", 0, 0, 0 ; e.g. chan1,2==irL, chan3,4==irR
> 
> then
> 
> adummy1, adummy2,ar3 , ar4		ftconv gaR, gir, ipartitionsize, 0, 0, 0 
> ar1 , ar2,adummy1,adummy2 		ftconv gaL, gir, ipartitionsize, 0, 0, 0 
> 
> this should preserve the relative amplitudes among the channels 
> 
> cheers,
> 
> Karin
> 
>> On 29 Jan 2016, at 06:35, akshay.rama  wrote:
>> 
>> Hi Steven,
>> 
>> I have attached the pconvolve code on the dropbox link too.
>> (https://www.dropbox.com/sh/q5hw3c3aa5jmvk0/AAAOne2ZeLTfV4snhiKV7p7qa?dl=0)
>> 
>> With the following code, the pconvolve function's audio output is perfectly
>> centred, but with the ftconv opcode, the stereo image is instantly shifted
>> slightly to the left. This is why I think the impulse response files are
>> still being scaled. 
>> 
>> Both IRL and IRR are stereo WAV files. 
>> 
>> The reason i shifted to the ftconv opcode is because I need to be able to
>> delete the wav impulse response files from the hard disk as soon as csound
>> initializes and loads the files. With pconv deleting the file causes an
>> error ( i think this is because it reads through the wav file every time the
>> filter is activated).
>> 
>> Thanks
>> 
>> 
>> 
>> --
>> View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746754.html
>> Sent from the Csound - General mailing list archive at Nabble.com.
>> 
>> Csound mailing list
>> Csound@listserv.heanet.ie
>> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
>> Send bugs reports to
>>       https://github.com/csound/csound/issues
>> Discussions of bugs and features can be posted here
> 
> Csound mailing list
> Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> Send bugs reports to
>        https://github.com/csound/csound/issues
> Discussions of bugs and features can be posted here

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

I looked at the code to pconvolve and ftconv (ftconv does some scaling
by calling csoundGetInverseRealFFTScale and multiplying, but that
function always returns 1.0, so it's not a factor). I also didn't see
anything significant between the pconvolve and ftconv examples.

On the other hand, I just realized looking at the screenshot that
Akshay is using Csound 5.17.  I have no idea if there was any changes
between then and now that might contribute to this (whether it's in
the gen01 code, or the convolution opcodes).

At this point, it'd be best to get the IR's to test.  Also, if Akshay
could try on Csound 6 that would give more information.




On Fri, Jan 29, 2016 at 4:42 AM, Victor Lazzarini
 wrote:
> That is a good idea, then you don’t even need to suppress rescaling. In any case, I just
> went and checkrf all the steps with a stereo IR I created, and I could not find any problems,
> the outputs of pconvolve and convolve are identical. Rescaling is suppressed as expected.
>
> This is the test. “ir.wav” is any stereo IR, and “input.wav” is any mono input file.
>
> nchnls=2
> 0dbfs=1
>
> gi1 ftgen 1,0,0,-1,"ir.wav",0,0,1
> gi2 ftgen 2,0,0,-1,"ir.wav”,0,0,2
>
> instr 1
> a1 diskin “input.wav"
> a2 ftconv a1,1,512
> a3 ftconv a1,2,512
>    outs a2,a3
> endin
> instr 2
> a1 diskin “input.wav"
> a2,a3 pconvolve a1,"ir.wav",512
>    outs a2,a3
> endin
>
> So whatever the problem is, it is not with rescaling, and not with ftconv.
>
> Regards
> ========================
> Dr Victor Lazzarini
> Dean of Arts, Celtic Studies and Philosophy,
> Maynooth University,
> Maynooth, Co Kildare, Ireland
> Tel: 00 353 7086936
> Fax: 00 353 1 7086952
>
>> On 29 Jan 2016, at 09:23, Karin Daum  wrote:
>>
>> another possibility I would try is combining the two stereo IR files into a single 4 channel IR file
>>
>> you would then use:
>>
>> gir ftgen 0, 0, 2048, -1, "IR4chwav", 0, 0, 0 ; e.g. chan1,2==irL, chan3,4==irR
>>
>> then
>>
>> adummy1, adummy2,ar3 , ar4            ftconv gaR, gir, ipartitionsize, 0, 0, 0
>> ar1 , ar2,adummy1,adummy2             ftconv gaL, gir, ipartitionsize, 0, 0, 0
>>
>> this should preserve the relative amplitudes among the channels
>>
>> cheers,
>>
>> Karin
>>
>>> On 29 Jan 2016, at 06:35, akshay.rama  wrote:
>>>
>>> Hi Steven,
>>>
>>> I have attached the pconvolve code on the dropbox link too.
>>> (https://www.dropbox.com/sh/q5hw3c3aa5jmvk0/AAAOne2ZeLTfV4snhiKV7p7qa?dl=0)
>>>
>>> With the following code, the pconvolve function's audio output is perfectly
>>> centred, but with the ftconv opcode, the stereo image is instantly shifted
>>> slightly to the left. This is why I think the impulse response files are
>>> still being scaled.
>>>
>>> Both IRL and IRR are stereo WAV files.
>>>
>>> The reason i shifted to the ftconv opcode is because I need to be able to
>>> delete the wav impulse response files from the hard disk as soon as csound
>>> initializes and loads the files. With pconv deleting the file causes an
>>> error ( i think this is because it reads through the wav file every time the
>>> filter is activated).
>>>
>>> Thanks
>>>
>>>
>>>
>>> --
>>> View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746754.html
>>> Sent from the Csound - General mailing list archive at Nabble.com.
>>>
>>> Csound mailing list
>>> Csound@listserv.heanet.ie
>>> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
>>> Send bugs reports to
>>>       https://github.com/csound/csound/issues
>>> Discussions of bugs and features can be posted here
>>
>> Csound mailing list
>> Csound@listserv.heanet.ie
>> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
>> Send bugs reports to
>>        https://github.com/csound/csound/issues
>> Discussions of bugs and features can be posted here
>
> Csound mailing list
> Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> Send bugs reports to
>         https://github.com/csound/csound/issues
> Discussions of bugs and features can be posted here

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

Very difficult to say anything about 5.07.
========================
Dr Victor Lazzarini
Dean of Arts, Celtic Studies and Philosophy,
Maynooth University,
Maynooth, Co Kildare, Ireland
Tel: 00 353 7086936
Fax: 00 353 1 7086952 

> On 29 Jan 2016, at 18:06, Steven Yi  wrote:
> 
> I looked at the code to pconvolve and ftconv (ftconv does some scaling
> by calling csoundGetInverseRealFFTScale and multiplying, but that
> function always returns 1.0, so it's not a factor). I also didn't see
> anything significant between the pconvolve and ftconv examples.
> 
> On the other hand, I just realized looking at the screenshot that
> Akshay is using Csound 5.17.  I have no idea if there was any changes
> between then and now that might contribute to this (whether it's in
> the gen01 code, or the convolution opcodes).
> 
> At this point, it'd be best to get the IR's to test.  Also, if Akshay
> could try on Csound 6 that would give more information.
> 
> 
> 
> 
> On Fri, Jan 29, 2016 at 4:42 AM, Victor Lazzarini
>  wrote:
>> That is a good idea, then you don’t even need to suppress rescaling. In any case, I just
>> went and checkrf all the steps with a stereo IR I created, and I could not find any problems,
>> the outputs of pconvolve and convolve are identical. Rescaling is suppressed as expected.
>> 
>> This is the test. “ir.wav” is any stereo IR, and “input.wav” is any mono input file.
>> 
>> nchnls=2
>> 0dbfs=1
>> 
>> gi1 ftgen 1,0,0,-1,"ir.wav",0,0,1
>> gi2 ftgen 2,0,0,-1,"ir.wav”,0,0,2
>> 
>> instr 1
>> a1 diskin “input.wav"
>> a2 ftconv a1,1,512
>> a3 ftconv a1,2,512
>>   outs a2,a3
>> endin
>> instr 2
>> a1 diskin “input.wav"
>> a2,a3 pconvolve a1,"ir.wav",512
>>   outs a2,a3
>> endin
>> 
>> So whatever the problem is, it is not with rescaling, and not with ftconv.
>> 
>> Regards
>> ========================
>> Dr Victor Lazzarini
>> Dean of Arts, Celtic Studies and Philosophy,
>> Maynooth University,
>> Maynooth, Co Kildare, Ireland
>> Tel: 00 353 7086936
>> Fax: 00 353 1 7086952
>> 
>>> On 29 Jan 2016, at 09:23, Karin Daum  wrote:
>>> 
>>> another possibility I would try is combining the two stereo IR files into a single 4 channel IR file
>>> 
>>> you would then use:
>>> 
>>> gir ftgen 0, 0, 2048, -1, "IR4chwav", 0, 0, 0 ; e.g. chan1,2==irL, chan3,4==irR
>>> 
>>> then
>>> 
>>> adummy1, adummy2,ar3 , ar4            ftconv gaR, gir, ipartitionsize, 0, 0, 0
>>> ar1 , ar2,adummy1,adummy2             ftconv gaL, gir, ipartitionsize, 0, 0, 0
>>> 
>>> this should preserve the relative amplitudes among the channels
>>> 
>>> cheers,
>>> 
>>> Karin
>>> 
>>>> On 29 Jan 2016, at 06:35, akshay.rama  wrote:
>>>> 
>>>> Hi Steven,
>>>> 
>>>> I have attached the pconvolve code on the dropbox link too.
>>>> (https://www.dropbox.com/sh/q5hw3c3aa5jmvk0/AAAOne2ZeLTfV4snhiKV7p7qa?dl=0)
>>>> 
>>>> With the following code, the pconvolve function's audio output is perfectly
>>>> centred, but with the ftconv opcode, the stereo image is instantly shifted
>>>> slightly to the left. This is why I think the impulse response files are
>>>> still being scaled.
>>>> 
>>>> Both IRL and IRR are stereo WAV files.
>>>> 
>>>> The reason i shifted to the ftconv opcode is because I need to be able to
>>>> delete the wav impulse response files from the hard disk as soon as csound
>>>> initializes and loads the files. With pconv deleting the file causes an
>>>> error ( i think this is because it reads through the wav file every time the
>>>> filter is activated).
>>>> 
>>>> Thanks
>>>> 
>>>> 
>>>> 
>>>> --
>>>> View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746754.html
>>>> Sent from the Csound - General mailing list archive at Nabble.com.
>>>> 
>>>> Csound mailing list
>>>> Csound@listserv.heanet.ie
>>>> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
>>>> Send bugs reports to
>>>>      https://github.com/csound/csound/issues
>>>> Discussions of bugs and features can be posted here
>>> 
>>> Csound mailing list
>>> Csound@listserv.heanet.ie
>>> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
>>> Send bugs reports to
>>>       https://github.com/csound/csound/issues
>>> Discussions of bugs and features can be posted here
>> 
>> Csound mailing list
>> Csound@listserv.heanet.ie
>> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
>> Send bugs reports to
>>        https://github.com/csound/csound/issues
>> Discussions of bugs and features can be posted here
> 
> Csound mailing list
> Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> Send bugs reports to
>        https://github.com/csound/csound/issues
> Discussions of bugs and features can be posted here

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

I can't provide the IR files since they are intellectual property. For the
past few days I have been trying the same code on my mac. On my mac, there
seems to be no difference in the audio output and there is no scaling.
However when running it on the Beaglebone Black the same problem persists. I
have upgraded to csound 6.07 on the beaglebone, but there still seems to be
apparent scaling. I have attached a screenshot of the printed output in the
same dropbox link. 

I have also tried changing the order of the convolution to see if it
affected the audio output and was the reason the stereo image seems shifting
to the left, however this has no affect in the audio output. 



--
View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746883.html
Sent from the Csound - General mailing list archive at Nabble.com.

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

Hi Karin,

This seems like a great way to ensure none of the stereo ir files are scaled
differently. I have been trying to implement this, but after loading the 4
channel IR file using : ir4 ftgen 0, 0, 2048, -1, "IR4ch.wav", 0, 0, 0 

I cant seem to find a way to split it into two different variables ( chan1,2
== irl , chan3,4 == irr ) before using it for the convolution. 

Thanks,





--
View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746884.html
Sent from the Csound - General mailing list archive at Nabble.com.

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

Well, can you provide any examples of the issue with other files than the copyrighted ones? 
We can’t reproduce what you describe with any files here. I showed an example here that
demonstrated that scaling is suppressed correctly and that pconvolve and ftconv behave
identically.

If it is working on the Mac and not on the Beaglebone, maybe there is a hardware 
issue.

========================
Dr Victor Lazzarini
Dean of Arts, Celtic Studies and Philosophy,
Maynooth University,
Maynooth, Co Kildare, Ireland
Tel: 00 353 7086936
Fax: 00 353 1 7086952 

> On 2 Feb 2016, at 07:20, akshay.rama  wrote:
> 
> I can't provide the IR files since they are intellectual property. For the
> past few days I have been trying the same code on my mac. On my mac, there
> seems to be no difference in the audio output and there is no scaling.
> However when running it on the Beaglebone Black the same problem persists. I
> have upgraded to csound 6.07 on the beaglebone, but there still seems to be
> apparent scaling. I have attached a screenshot of the printed output in the
> same dropbox link. 
> 
> I have also tried changing the order of the convolution to see if it
> affected the audio output and was the reason the stereo image seems shifting
> to the left, however this has no affect in the audio output. 
> 
> 
> 
> --
> View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746883.html
> Sent from the Csound - General mailing list archive at Nabble.com.
> 
> Csound mailing list
> Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> Send bugs reports to
>        https://github.com/csound/csound/issues
> Discussions of bugs and features can be posted here

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

well, I’m not an expert on Csound. I’m just starting and convolution with pulse response files (surround 5.1 in my case) was the first thing I looked at. My proposal was just tailored to your example. I’m afraid there is no simple way by means of a single opcode to extract the two channels to another function table directly. I guess you would need to do it by hand using table and then writing it to an empty table via tablew. I have just a similar problem having stored a couple of small pieces of sound in tables which later-on have to be processed.

Concerning pulse response files Victor asked for, I’ve attached two pulse response files I’ve calculated for two different source position in my setup ( a virtual room of 20 x 50 metres high reflectivity of 0.87 resulting in a hall time T_60 of about 4 second - something like a cathedral) for 5.1 and then converted to stereo. These can be used for tests.

cheers,

Karin

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here


> On 2 Feb 2016, at 08:24, akshay.rama  wrote:
> 
> Hi Karin,
> 
> This seems like a great way to ensure none of the stereo ir files are scaled
> differently. I have been trying to implement this, but after loading the 4
> channel IR file using : ir4 ftgen 0, 0, 2048, -1, "IR4ch.wav", 0, 0, 0 
> 
> I cant seem to find a way to split it into two different variables ( chan1,2
> == irl , chan3,4 == irr ) before using it for the convolution. 
> 
> Thanks,
> 
> 
> 
> 
> 
> --
> View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746884.html
> Sent from the Csound - General mailing list archive at Nabble.com.
> 
> Csound mailing list
> Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> Send bugs reports to
>        https://github.com/csound/csound/issues
> Discussions of bugs and features can be posted here


Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

I forgot to say that the channels are stored for every sample t as ch1,ch2,ch3,ch4 so for a given ksmps you would need to read 4*ksmps elements from the table in the order ch1,ch2,ch3,ch4,ch1,ch2,ch3,ch4,…
cheers,

Karin

> On 2 Feb 2016, at 08:24, akshay.rama  wrote:
> 
> Hi Karin,
> 
> This seems like a great way to ensure none of the stereo ir files are scaled
> differently. I have been trying to implement this, but after loading the 4
> channel IR file using : ir4 ftgen 0, 0, 2048, -1, "IR4ch.wav", 0, 0, 0 
> 
> I cant seem to find a way to split it into two different variables ( chan1,2
> == irl , chan3,4 == irr ) before using it for the convolution. 
> 
> Thanks,
> 
> 
> 
> 
> 
> --
> View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746884.html
> Sent from the Csound - General mailing list archive at Nabble.com.
> 
> Csound mailing list
> Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> Send bugs reports to
>        https://github.com/csound/csound/issues
> Discussions of bugs and features can be posted here

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

Thanks for the files, but I need someone to provide an example showing the rescaling is not happening (or that there is some other fault).
========================
Dr Victor Lazzarini
Dean of Arts, Celtic Studies and Philosophy,
Maynooth University,
Maynooth, Co Kildare, Ireland
Tel: 00 353 7086936
Fax: 00 353 1 7086952 

> On 2 Feb 2016, at 10:05, Karin Daum  wrote:
> 
> well, I’m not an expert on Csound. I’m just starting and convolution with pulse response files (surround 5.1 in my case) was the first thing I looked at. My proposal was just tailored to your example. I’m afraid there is no simple way by means of a single opcode to extract the two channels to another function table directly. I guess you would need to do it by hand using table and then writing it to an empty table via tablew. I have just a similar problem having stored a couple of small pieces of sound in tables which later-on have to be processed.
> 
> Concerning pulse response files Victor asked for, I’ve attached two pulse response files I’ve calculated for two different source position in my setup ( a virtual room of 20 x 50 metres high reflectivity of 0.87 resulting in a hall time T_60 of about 4 second - something like a cathedral) for 5.1 and then converted to stereo. These can be used for tests.
> 
> cheers,
> 
> Karin
> 
> Csound mailing list
> Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> Send bugs reports to
>        https://github.com/csound/csound/issues
> Discussions of bugs and features can be posted here
> 
> 
>> On 2 Feb 2016, at 08:24, akshay.rama  wrote:
>> 
>> Hi Karin,
>> 
>> This seems like a great way to ensure none of the stereo ir files are scaled
>> differently. I have been trying to implement this, but after loading the 4
>> channel IR file using : ir4 ftgen 0, 0, 2048, -1, "IR4ch.wav", 0, 0, 0 
>> 
>> I cant seem to find a way to split it into two different variables ( chan1,2
>> == irl , chan3,4 == irr ) before using it for the convolution. 
>> 
>> Thanks,
>> 
>> 
>> 
>> 
>> 
>> --
>> View this message in context: http://csound.1045644.n5.nabble.com/Csound-Function-Table-Scaling-Error-tp5746749p5746884.html
>> Sent from the Csound - General mailing list archive at Nabble.com.
>> 
>> Csound mailing list
>> Csound@listserv.heanet.ie
>> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
>> Send bugs reports to
>>       https://github.com/csound/csound/issues
>> Discussions of bugs and features can be posted here
> 
> 
> Csound mailing list
> Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> Send bugs reports to
>        https://github.com/csound/csound/issues
> Discussions of bugs and features can be posted here

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here