linuxsampler-devel

[Linuxsampler-devel] Gigedit: Positive Gain

[Christian S. <sch...@li...>]

Hi,

I just committed a change to Gigedit which allows fine graded positive gain 
values, and therefore I dropped the old "Gain +6dB" checkbox from Gigedit. So 
you can now set any gain value between -96dB..+96dB.

Before that change, only -96dB..0dB and +6dB were allowed, which was quite 
unpleasant since sometimes you just need to raise volume by e.g. +1.5dB or so 
which was so far not possible at all.

Does this break behaviour with the original GigaStudio player? Unless you 
retain the old allowed value range of -96..0,+6 then definitely no; because if 
you checked "+6 dB" before then Gigedit will simply show it now as 6.00 value 
and vice versa if you set the value to exactly 6.00 in Gigedit and save the 
.gig file then you get the exact same data on gig file level as if you were 
using the +6dB checkbox before.

However if you are using any new range value like e.g. +2.00 dB or +12.43dB? 
Well, I am not sure, but I am optimistic that even that should work as 
expected with Tascam's player, even though never supported by the original 
GigaStudio instrument editor. Because if you were checking "+6dB" before, that 
actually did not set a flag, it was always using a gain value on gig file 
level perfectly linear to any other possible -96dB..0dB value, which suggests 
that Tascam multiplied that value from the gig file with a constant factor to 
get the final floating point gain factor, exactly like we do in LS.

If I am wrong, you tell me. ;-)

CU
Christian

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Ivan M. <iv...@ma...>]

Hi Christian

I get the following behaviour in GSEdit3 using the new gain option:

Gain of -96 to 0dB in gigedit => Attenuation of 0 to 96dB, 6dB boost set 
to '0 - No' in GSEdit
Gain of 6dB in gigedit => Attenuation of 0dB, 6dB boost set to '1 - Yes' 
in GSEdit
Gain > than 0dB other than 6dB => Attenuation of the negative of the 
gain value, 6dB boost set to '0 - No' in GSEdit

There's a problem however, I'm not able to load any instruments created 
from scratch in gigedit as they cause GStudio3 to crash. In GSEdit3 I'm 
able to load the same instruments but I get the warning message "File is 
protected. Save As and Wave Export functions will be disabled." I 
believe the problem is that libgig is using a chunk for an LS extension 
which was already in use by GStudio3. Specifically, I think the culprit 
is the '3LS' tag and that the GigaStudio developers were planning to use 
this for a future copy protection feature. The 3LS tag was added at 
libgig revision 2584. Apologies for not spotting this sooner but I 
didn't really know how to use gigedit and GSEdit until earlier this year.

Let me know if there's anything else I can do.

All the best,
Ivan

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Andreas P. <and...@ou...>]

Ivan Maguidhir wrote:
> There's a problem however, I'm not able to load any instruments created
> from scratch in gigedit as they cause GStudio3 to crash. In GSEdit3 I'm
> able to load the same instruments but I get the warning message "File is
> protected. Save As and Wave Export functions will be disabled."

Hi,

I don't think the crash and the warning are related. We have always had
that warning, it's because we don't implement the encrypted checksum
GSEdit uses as some kind of protection.

/Andreas

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Christian S. <sch...@li...>]

On Sonntag, 8. Dezember 2019 08:31:48 CET Andreas Persson wrote:
> Ivan Maguidhir wrote:
> > There's a problem however, I'm not able to load any instruments created
> > from scratch in gigedit as they cause GStudio3 to crash. In GSEdit3 I'm
> > able to load the same instruments but I get the warning message "File is
> > protected. Save As and Wave Export functions will be disabled."
> 
> Hi,
> 
> I don't think the crash and the warning are related. We have always had
> that warning, it's because we don't implement the encrypted checksum
> GSEdit uses as some kind of protection.

I was not aware about that GSt crash before. Any suspicion about the cause of 
that GSt crash?

Ivan, you could check your assumption and replace the defines for our own 
chunk names, namely LIST_TYPE_3LS and LIST_TYPE_RTIS in src/gig.h by something 
else.

Since Andreas assumes the crash to be a different issue. You might also just 
copy any original GSt .gig file from some commercial source and just replace 
some gain value with gigedit there, since that would mean less changes to the 
gig file than creating one from scratch.

CU
Christian

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Andreas P. <and...@ou...>]

Christian Schoenebeck wrote:
> On Sonntag, 8. Dezember 2019 08:31:48 CET Andreas Persson wrote:
>> Ivan Maguidhir wrote:
>>> There's a problem however, I'm not able to load any instruments created
>>> from scratch in gigedit as they cause GStudio3 to crash. In GSEdit3 I'm
>>> able to load the same instruments but I get the warning message "File is
>>> protected. Save As and Wave Export functions will be disabled."
>>
>> Hi,
>>
>> I don't think the crash and the warning are related. We have always had
>> that warning, it's because we don't implement the encrypted checksum
>> GSEdit uses as some kind of protection.
> 
> I was not aware about that GSt crash before. Any suspicion about the cause of 
> that GSt crash?

No, sorry, I don't know why it crashes.

> Since Andreas assumes the crash to be a different issue.

I didn't mean that the crash isn't related to the '3LS' tag, it might
very well be. I just meant that the warning message is not related to
the crash, as the warning has always been there.

/Andreas

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Ivan M. <iv...@ma...>]

Hi Christian

On 08/12/2019 09:51, Christian Schoenebeck wrote:
> Ivan, you could check your assumption and replace the defines for our own
> chunk names, namely LIST_TYPE_3LS and LIST_TYPE_RTIS in src/gig.h by something
> else.

I think Andreas is right. On modifying the chunk IDs for LIST_TYPE_3LS 
and LIST_TYPE_RTIS I get the same warning about the file being 
protected. I also now get the message: "This file has an irregular 
structure. One possible explanation is that it was written by an 
application other than the GigaStudio Editor. A full save will be 
required if you modify the file." but I think this is probably something 
to do with the new chunk ID values I chose. can

> Since Andreas assumes the crash to be a different issue. You might also just
> copy any original GSt .gig file from some commercial source and just replace
> some gain value with gigedit there, since that would mean less changes to the
> gig file than creating one from scratch.

I made two copies of a commercial instrument and set the gain to +6dB 
and +12dB respectively using gigedit. Both copies load correctly in 
GSEdit3 and GStudio3. The +6dB instrument is louder than the original, 
but the +12dB instrument seems to be the same volume as the original 
instrument. I'll do a more scientific test later and record the audio 
output.

I didn't use Save As in gigedit to create the copies above as 
instruments I created with it caused GStudio3 to crash.

> CU
> Christian

All the best,
Ivan

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Christian S. <sch...@li...>]

On Sonntag, 8. Dezember 2019 19:22:20 CET Ivan Maguidhir wrote:
> Hi Christian
> 
> On 08/12/2019 09:51, Christian Schoenebeck wrote:
> > Ivan, you could check your assumption and replace the defines for our own
> > chunk names, namely LIST_TYPE_3LS and LIST_TYPE_RTIS in src/gig.h by
> > something else.
> 
> I think Andreas is right. On modifying the chunk IDs for LIST_TYPE_3LS
> and LIST_TYPE_RTIS I get the same warning about the file being
> protected. I also now get the message: "This file has an irregular
> structure. One possible explanation is that it was written by an
> application other than the GigaStudio Editor. A full save will be
> required if you modify the file." but I think this is probably something
> to do with the new chunk ID values I chose. can

Does that mean we can leave the current chunk names for our gig format 
extensions? So after you changed the chunk names you got the exact same 
behaviour as yesterday?

I mean, I just added these 2 chunk types just 2 months ago, so now is still an 
appropriate time to rename them if required. Later on (i.e. after the next 
official release) that would become more complicated and nasty.

BTW those 2 custom RIFF chunks should actually just be created by libgig if 
you really used those custom features in Gigedit.

> > Since Andreas assumes the crash to be a different issue. You might also
> > just copy any original GSt .gig file from some commercial source and just
> > replace some gain value with gigedit there, since that would mean less
> > changes to the gig file than creating one from scratch.
> 
> I made two copies of a commercial instrument and set the gain to +6dB
> and +12dB respectively using gigedit. Both copies load correctly in
> GSEdit3 and GStudio3. The +6dB instrument is louder than the original,
> but the +12dB instrument seems to be the same volume as the original
> instrument. I'll do a more scientific test later and record the audio
> output.
> 
> I didn't use Save As in gigedit to create the copies above as
> instruments I created with it caused GStudio3 to crash.

And that's the same GSt crash you mentioned yesterday? Or do you talk about 
multiple different GSt crashes?

CU
Christian

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Ivan M. <iv...@ma...>]

Hi Christian

On 08/12/2019 19:05, Christian Schoenebeck wrote:
>
> Does that mean we can leave the current chunk names for our gig format
> extensions? So after you changed the chunk names you got the exact same
> behaviour as yesterday?

Yes. GSEdit3 displays the 'protected' message even when I create 
instruments with alternative values for the LIST_TYPE_3LS and 
LIST_TYPE_RTIS chunk IDs.

Also, the message about 'irregular structure' I mentioned earlier isn't 
displayed when Loops and Filters are both enabled and the instrument 
name is set (in Instrument Properties) in gigedit.

> BTW those 2 custom RIFF chunks should actually just be created by libgig if
> you really used those custom features in Gigedit.
Yes I know. They weren't actually included in the instruments I created 
yesterday. The '3LS' I spotted in my hex editor was part of the sample data.
>
> And that's the same GSt crash you mentioned yesterday? Or do you talk about
> multiple different GSt crashes?
>
The crash has nothing to do with using Save As as I thought earlier. 
Instruments created from scratch with gigedit cause a crash. Commercial 
and self-made instruments created with GSEdit and re-saved with gigedit 
don't cause it.

Oddly, if I load a gigedit authored instrument in GSEdit3 and exit the 
program without making any changes it asks me if I want to 'Save 
Changes'. If I click Yes the saved instrument doesn't crash GSt3 but 
also doesn't produce sound as intended.


> CU
> Christian

All the best,
Ivan

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Ivan M. <iv...@ma...>]

Hi Christian

I made multiple copies of a commercial gig instrument with gigedit 
setting the gain option to -3dB, 0dB, +3dB, +6dB, +12dB. I've attached 
the audio output of GSt3 when the same MIDI note is played on each 
instrument in succession. I only realised after doing this that the 
instrument I used is a version 1.0 gig. I'm not sure if this would make 
a difference? I can redo it with a v3 instrument if necessary.

All the best,
Ivan

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Christian S. <sch...@li...>]

On Montag, 9. Dezember 2019 01:53:05 CET Ivan Maguidhir wrote:
> Also, the message about 'irregular structure' I mentioned earlier isn't
> displayed when Loops and Filters are both enabled and the instrument
> name is set (in Instrument Properties) in gigedit.

Hmmm... I can't see the connection.

The info you can enter by the Instruments Properties dialog is written to DLS 
Info chunks, and these DLS Info chunks are always written, no matter if you 
entered something or not. The instrument name field (if not provided) is then 
filled with "NONAME" by default (see DLS.cpp).

The filter enabled setting is just a bit flag on gig file level (see 
VCFEnabled in gig.cpp). So that filter setting does not have any influence on 
any data structure.

Maybe it's really just the loop setting that made the difference there. 
Because that indeed causes a data structure change.

CU
Christian

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Ivan M. <iv...@ma...>]

Hi Christian

On 10/12/2019 10:25, Christian Schoenebeck wrote:
>
> Maybe it's really just the loop setting that made the difference there.
> Because that indeed causes a data structure change.
Yes, you're right. I created an instrument from scratch again with 
gigedit. Without Loops enabled I get the 'irregular structure' message 
in GSEdit3 and with Loops enabled the message doesn't appear. Using 
GSEdit3 to re-save the instrument solves the problem also.
> CU
> Christian
>
All the best,
Ivan

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Christian S. <sch...@li...>]

On Montag, 9. Dezember 2019 22:24:49 CET Ivan Maguidhir wrote:
> Hi Christian
> 
> I made multiple copies of a commercial gig instrument with gigedit
> setting the gain option to -3dB, 0dB, +3dB, +6dB, +12dB. I've attached
> the audio output of GSt3 when the same MIDI note is played on each
> instrument in succession.

Not really the result I would expect. Looking at your audio file 1st note is 
-3dB (ok), 2nd note is 0dB (ok), but then 3rd is also 0dB (instead of +3dB, 
maybe ok because its not allowed by GSt), then 4th is -3dB instead of +6dB 
which is definitely wrong because that's a setting supported by GSt, and well 
5th note is again 0dB instead of +12dB (but that's again an unsupported option 
in GSt).

Just to make this clear about deciBels in general: 0dB means "neutral" / no 
volume change, +6dB means increasing volume by factor 2, -6dB means half 
volume, +12dB means volume increase by factor 4, -12dB means volume reduced to 
1/4th. So you can easily see those volume relations already with a wave 
editor.

Did you check "Apply to:" ... "All regions" checkbox and "All Dimension 
Regions" checkbox in Gigedit before you did these gain changes?

> I only realised after doing this that the
> instrument I used is a version 1.0 gig. I'm not sure if this would make
> a difference? I can redo it with a v3 instrument if necessary.

Not sure either, but this gain setting is fairly old and already existed 
including that +6dB option when I wrote the first line of libgig code in 2003.

CU
Christian

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Ivan M. <iv...@ma...>]

Hi Christian

On 10/12/2019 12:06, Christian Schoenebeck wrote:
>
> Did you check "Apply to:" ... "All regions" checkbox and "All Dimension
> Regions" checkbox in Gigedit before you did these gain changes?

Yes, I had applied the gain setting to all regions of each instrument in 
gigedit. I loaded the instruments one by one in GSt3 again just now and 
I got the attached result. I must have loaded the same instrument in 
multiple channels yesterday as I didn't modify the instruments in the 
interim.


> CU
> Christian

All the best,
Ivan

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Christian S. <sch...@li...>]

On Dienstag, 10. Dezember 2019 19:04:01 CET Ivan Maguidhir wrote:
> Hi Christian
> 
> On 10/12/2019 12:06, Christian Schoenebeck wrote:
> > Did you check "Apply to:" ... "All regions" checkbox and "All Dimension
> > Regions" checkbox in Gigedit before you did these gain changes?
> 
> Yes, I had applied the gain setting to all regions of each instrument in
> gigedit. I loaded the instruments one by one in GSt3 again just now and
> I got the attached result. I must have loaded the same instrument in
> multiple channels yesterday as I didn't modify the instruments in the
> interim.

Now that looks much better! :) So the old allowed gain settings look all fine 
with GSt, and GSt player even supports the +3d gain setting correctly (even 
though never supported by GSt's instrument editor). Only the +12dB gain 
setting seems to be ignored by GSt and it used +6dB instead. So either GSt 
player limits the gain setting to max. +6dB or there was some compression 
going on. But it rather looks like gain parameter limit of +6dB on first 
sight.

Anyway, looks good to me! Thanks!

CU
Christian

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Christian S. <sch...@li...>]

On Dienstag, 10. Dezember 2019 20:41:54 CET Ivan Maguidhir wrote:
> On 10/12/2019 10:25, Christian Schoenebeck wrote:
> > Maybe it's really just the loop setting that made the difference there.
> > Because that indeed causes a data structure change.
> 
> Yes, you're right. I created an instrument from scratch again with
> gigedit. Without Loops enabled I get the 'irregular structure' message
> in GSEdit3 and with Loops enabled the message doesn't appear. Using
> GSEdit3 to re-save the instrument solves the problem also.

Hmm... I wonder what GSt actually expects on gig file level if no loop was 
enabled. I mean it's probably not a big deal if GSt spits out a warning on 
such gig files, but would be nice to have if fixed on libgig side if 
reasonable.

Can you identify what GSt instrument editor changes/fixes after you created a 
simple .gig file from scratch with Gigedit with no loop on?

CU
Christian

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Ivan M. <iv...@ma...>]

Hi Christian

On 11/12/2019 12:30, Christian Schoenebeck wrote:
>
> Hmm... I wonder what GSt actually expects on gig file level if no loop was
> enabled. I mean it's probably not a big deal if GSt spits out a warning on
> such gig files, but would be nice to have if fixed on libgig side if
> reasonable.
>
> Can you identify what GSt instrument editor changes/fixes after you created a
> simple .gig file from scratch with Gigedit with no loop on?

Yes, I managed to get GSEdit3 to output debug logs showing the structure 
(as it sees it) of input gig files (like gigdump in LS). This showed me 
several differences between gig files created with gigedit and the ones 
corrected by GSEdit3. The problem causing the 'irregular structure' 
message was the wrong number of '3gnm' strings - there should always be 128.

There was also a duplicate DimensionRegion chunk being created when 
designing an instrument from scratch with gigedit.

Finally, I discovered two new chunks from the official spec. They always 
seem to have the same content and I have no idea what they do :-) I just 
wanted to rule them out as a cause of the problem.

I've attached a patch with these changes.

>
> CU
> Christian
>
All the best,
Ivan

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Christian S. <sch...@li...>]

On Donnerstag, 12. Dezember 2019 04:38:45 CET Ivan Maguidhir wrote:
> Hi Christian
> 
> On 11/12/2019 12:30, Christian Schoenebeck wrote:
> > Hmm... I wonder what GSt actually expects on gig file level if no loop was
> > enabled. I mean it's probably not a big deal if GSt spits out a warning on
> > such gig files, but would be nice to have if fixed on libgig side if
> > reasonable.
> > 
> > Can you identify what GSt instrument editor changes/fixes after you
> > created a simple .gig file from scratch with Gigedit with no loop on?
> 
> Yes, I managed to get GSEdit3 to output debug logs showing the structure
> (as it sees it) of input gig files (like gigdump in LS). This showed me
> several differences between gig files created with gigedit and the ones
> corrected by GSEdit3. The problem causing the 'irregular structure'
> message was the wrong number of '3gnm' strings - there should always be 128.
> 
> There was also a duplicate DimensionRegion chunk being created when
> designing an instrument from scratch with gigedit.
> 
> Finally, I discovered two new chunks from the official spec. They always
> seem to have the same content and I have no idea what they do :-) I just
> wanted to rule them out as a cause of the problem.
> 
> I've attached a patch with these changes.

Nice job! :)

Give me some time, I probably have to make some minor adjustments to your 
patch.

Thanks!

CU
Christian

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Ivan M. <iv...@ma...>]

On 13/12/2019 13:40, Christian Schoenebeck wrote:
>
> Nice job! :)
>
> Give me some time, I probably have to make some minor adjustments to your
> patch.
>
> Thanks!
No problem. I'm glad it's helpful. I'll take a look at the GSt3 crash 
that happens with newly created gigedit instruments next.
> CU
> Christian
>
All the best,
Ivan

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Christian S. <sch...@li...>]

On Freitag, 13. Dezember 2019 18:13:00 CET Ivan Maguidhir wrote:
> On 13/12/2019 13:40, Christian Schoenebeck wrote:
> > Nice job! :)
> > 
> > Give me some time, I probably have to make some minor adjustments to your
> > patch.

Ok, I just committed 2 parts of your patch. I committed your 128 '3gnm' chunks 
patch unmodified:
http://svn.linuxsampler.org/cgi-bin/viewvc.cgi?view=revision&revision=3656

Regarding the '3dnm' and '3ddp' chunks: I moved that to UpdateChunks() since I 
found that the more appropriate place to be, and I assumed that these chunks 
are just required for GigaStudio >= v3, hence these are only added there and 
auto removed if user switches back to a previous gig format version:
http://svn.linuxsampler.org/cgi-bin/viewvc.cgi?view=revision&revision=3657

Then your changes to AddDimension(): I could not reproduce that issue and 
accordingly I haven't committed that part of your patch yet. In my tests, when 
I start with an Instrument with no dimension and then add another dimension I 
never ended up with duplicate RIFF chunks. Plus, as far as I can see it, your 
suggested changes would introduce other bugs, because e.g. you would be 
assigning the same DimensionRegion object several times to different 
'pDimensionRegions' array entries. So are you sure you still have issues with 
latest libgig SVN version that we would need to address here?

CU
Christian

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Ivan M. <iv...@ma...>]

Hi Christian

On 14/12/2019 17:26, Christian Schoenebeck wrote:
> Ok, I just committed 2 parts of your patch. I committed your 128 '3gnm' chunks
> patch unmodified:
> http://svn.linuxsampler.org/cgi-bin/viewvc.cgi?view=revision&revision=3656
Thanks!
> Regarding the '3dnm' and '3ddp' chunks: I moved that to UpdateChunks() since I
> found that the more appropriate place to be, and I assumed that these chunks
> are just required for GigaStudio >= v3, hence these are only added there and
> auto removed if user switches back to a previous gig format version:
> http://svn.linuxsampler.org/cgi-bin/viewvc.cgi?view=revision&revision=3657
That sounds good. Actually, I found the '3ddp' tag in several v1.0 
instruments that I have. I'll have another look at GSEdit3 later today 
to see if I can find out exactly what both of these tags do.
>
> Then your changes to AddDimension(): I could not reproduce that issue and
> accordingly I haven't committed that part of your patch yet. In my tests, when
> I start with an Instrument with no dimension and then add another dimension I
> never ended up with duplicate RIFF chunks. Plus, as far as I can see it, your
> suggested changes would introduce other bugs, because e.g. you would be
> assigning the same DimensionRegion object several times to different
> 'pDimensionRegions' array entries. So are you sure you still have issues with
> latest libgig SVN version that we would need to address here?

That's weird. I'm not seeing any duplication of the DimensionRegion 
chunks with the latest libgig from SVN.
So, you can disregard the last part of the patch. Sorry about that.

>
> CU
> Christian
>
>
All the best,
Ivan

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Christian S. <sch...@li...>]

On Samstag, 14. Dezember 2019 20:05:35 CET Ivan Maguidhir wrote:
> > Regarding the '3dnm' and '3ddp' chunks: I moved that to UpdateChunks()
> > since I found that the more appropriate place to be, and I assumed that
> > these chunks are just required for GigaStudio >= v3, hence these are only
> > added there and auto removed if user switches back to a previous gig
> > format version:
> > http://svn.linuxsampler.org/cgi-bin/viewvc.cgi?view=revision&revision=365
> > 7
> That sounds good. Actually, I found the '3ddp' tag in several v1.0
> instruments that I have. I'll have another look at GSEdit3 later today
> to see if I can find out exactly what both of these tags do.

In gig v1 even. I can't remember having seen those chunks before with GSt < 3 
files. But if it really turns out that these chunks were always there, then we 
can still change that of course and add them for all gig file format versions.

In my TODO list back from that time (2003) I only had a '3dnl' chunk not being 
supported by libgig on Region level yet:
http://svn.linuxsampler.org/cgi-bin/viewvc.cgi/libgig/trunk/TODO?view=markup

> > Then your changes to AddDimension(): I could not reproduce that issue and
> > accordingly I haven't committed that part of your patch yet. In my tests,
> > when I start with an Instrument with no dimension and then add another
> > dimension I never ended up with duplicate RIFF chunks. Plus, as far as I
> > can see it, your suggested changes would introduce other bugs, because
> > e.g. you would be assigning the same DimensionRegion object several times
> > to different 'pDimensionRegions' array entries. So are you sure you still
> > have issues with latest libgig SVN version that we would need to address
> > here?
> 
> That's weird. I'm not seeing any duplication of the DimensionRegion
> chunks with the latest libgig from SVN.
> So, you can disregard the last part of the patch. Sorry about that.

Maybe you were dropping a stereo sample in and misinterpreted the 
automatically created stereo dimension (a.k.a. "audio channel" dimension) as 
duplicate chunks.

At least so far I couldn't reproduce a problem regarding duplicate chunks 
there, so I leave it as is for now, but I'll keep an eye on it.

CU
Christian

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Andreas P. <and...@ou...>]

Christian Schoenebeck wrote:
> In my TODO list back from that time (2003) I only had a '3dnl' chunk not being 
> supported by libgig on Region level yet:
> http://svn.linuxsampler.org/cgi-bin/viewvc.cgi/libgig/trunk/TODO?view=markup

I also had a look in an old text file of mine. There I listed 3dnm, 3dpp
and 3dnl under "things not needed to be fixed". No explanation, but I
guess they were not needed for the file to work in GSEdit.

Further notes from that file:

"3dnm/nam0, nam1, ...
0-31 - name of dimension
32 + 12*i - name of split

3dpp - most often filled with FF. Length 10 for gig v2, length 16 for v3.

3dnl - unknown empty list, only in gig v2. Maybe name of dimensions?"


/Andreas

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Ivan M. <iv...@ma...>]

On 15/12/2019 13:34, Christian Schoenebeck wrote:
> In gig v1 even. I can't remember having seen those chunks before with GSt < 3
> files. But if it really turns out that these chunks were always there, then we
> can still change that of course and add them for all gig file format versions.
>
> In my TODO list back from that time (2003) I only had a '3dnl' chunk not being
> supported by libgig on Region level yet:
> http://svn.linuxsampler.org/cgi-bin/viewvc.cgi/libgig/trunk/TODO?view=markup

The '3ddp' chunks are there in all of the XSample Set instruments from 
Best Service that I have. These were all created with GSEdit v1. Andreas 
is correct about the '3ddp' chunk's size, however it's also 10 bytes for 
gig v1.
'3dnl' is the Dimension Region name list chunk for gig v1 and gig v2. 
'3dnm' replaced '3dnl' as the Dimension Region name list chunk for gig v3.

Interestingly, the '3ddp' chunk is only written by GSEdit and never 
read. Which makes me think it's only of use to GSt when playing 
instruments for effects or the mixer (or something similar) as the 
'3ddp' values seem to be switches: 0x00 for off 0xFF for on. I'm about 
to test this theory now.

>
> Maybe you were dropping a stereo sample in and misinterpreted the
> automatically created stereo dimension (a.k.a. "audio channel" dimension) as
> duplicate chunks.

No it was a mono sample alright. Actually, i just realised that I still 
have the instrument. It was created on the 7th of December so that could 
explain we can't reproduce the same behaviour now. If it's useful I can 
send you the instrument as it's only 280k in size.

>
> CU
> Christian
>
All the best,
Ivan

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Ivan M. <iv...@ma...>]

On 15/12/2019 20:59, Ivan Maguidhir wrote:
>
>
> Interestingly, the '3ddp' chunk is only written by GSEdit and never 
> read. Which makes me think it's only of use to GSt when playing 
> instruments for effects or the mixer (or something similar) as the 
> '3ddp' values seem to be switches: 0x00 for off 0xFF for on. I'm about 
> to test this theory now.
>
>
I checked this and the '3ddp' chunk isn't read by GSt either. It doesn't 
look like it's used at all.

All the best,
Ivan

Re: [Linuxsampler-devel] Gigedit: Positive Gain

[Christian S. <sch...@li...>]

On Sonntag, 15. Dezember 2019 21:59:37 CET Ivan Maguidhir wrote:
> No it was a mono sample alright. Actually, i just realised that I still
> have the instrument. It was created on the 7th of December so that could
> explain we can't reproduce the same behaviour now. If it's useful I can
> send you the instrument as it's only 280k in size.

Let's postpone that. I won't have the time to check that file soon. We'll keep 
an eye on this issue, and if it really happens again we'll take closer look.

> > Interestingly, the '3ddp' chunk is only written by GSEdit and never
> > read. Which makes me think it's only of use to GSt when playing
> > instruments for effects or the mixer (or something similar) as the
> > '3ddp' values seem to be switches: 0x00 for off 0xFF for on. I'm about
> > to test this theory now.
> 
> I checked this and the '3ddp' chunk isn't read by GSt either. It doesn't
> look like it's used at all.

So what do we do about it, let's drop '3ddp' from libgig code again then?

Best regards,
Christian Schoenebeck