Re: [Linuxsampler-devel] Re: EG

[Vladimir S. <ha...@so...>]

Christian,

Cool!
Nice diagram. Way better than my ascii graph :)

One little thing i noticed is that it doesn't go from attack_hold into 
release.
I had the same question at some point and i found a few gigs with 
attackhold=1.
I've played them in Gst and it certainly goes from attack_hold into release.
've used a short script and libgig to find gigs to try:
#!/bin/bash
#
for gig in *.[gG][iI][Gg]; do
#echo lookin at gig: $gig
ishold=`gigdump $gig|grep "EG1Hold=1"`
if [ "$ishold" != "" ]; then
        echo "Found gig with EG1Hold=1: $gig"
fi
done

Basically none of the gigs i found played correctly until i introduced a 
transition from attack_hold into release. I'm not sure if that's 
actually correct or not but GSt does it so i'm assuming it is.

Hope this helps!
Regards,
Vladimir.

PS: My jack seems to go crazy only after i run out of voices. It then 
continues to play "out of tune" until i run out of voices again at that 
point it chanes tune again and sometimes recovers sometimes becomes even 
more out of tune . . . Eventually after i run out of voices a few times 
in a row it recovers :)

PPS: Sorry i didn't pay attention to the jitter stuff. I thought it was 
just a small merge and didn't reallize it had to be implemented for all 
stages.

Christian Schoenebeck wrote:

>Es geschah am Sonntag, 8. Februar 2004 03:51 als Vladimir Senkov schrieb:
>  
>
>>Hi Christian,
>>
>>Here is the cumulative patch against latest CVS.
>>It includes:
>>1) ADSDR implementaion (includes a hack for "post release sustain")
>>    
>>
>
>I had a look at your patch, but it wasn't applicable, sorry. Especially due to 
>the lack of event handling in the new states I chose another way. The problem 
>is, it would introduce jitter. You might think this is not that important, 
>but think about the "freeze" scenario where you might have very big fragment 
>sizes. Anyway, this is the state diagram of the new envelope generator 
>(already in CVS):
>
>	http://www.linuxsampler.org/doc/engines/gig/eg1.pdf
>
>Please have a look at it (Vladimir, Mark) and let me know if you think it's 
>incorrect compared to the EG from Gst.
>
>  
>
>>2) Command line options (alsa card selection, jack playback autoconnect
>>and alsa samplerate)
>>    
>>
>
>I applied that to CVS, I only did some minor cosmetic adjustment.
>
>  
>
>>3) I've fixed EOL characters in jackio.h
>>    
>>
>
>I fixed that manually on the CVS server today for all files that had that 
>problem, so that CVS history keeps consistent.
>
>  
>
>>Regards,
>>Vladimir.
>>    
>>
>
>Thanks Vladimir!
>
>  
>
>>PS: I've been playing with jack for a little while. Something strange is
>>going on. Sometimes, (maybe under load?) LS goes out of tune. It begins
>>to sound as if i had my card at 32KHz but LS was thinking it was at 48 . .
>>. It stays that way until i restart it. It never happened with alsa though.
>>I haven't been able to consistently reproduce this yet. Have you had
>>anything like this happening at all?
>>    
>>
>
>I never had that problem nor did I hear something about that. Anybody else 
>having this problem?
>
>CU
>Christian
>
>
>
>-------------------------------------------------------
>SF.Net is sponsored by: Speed Start Your Linux Apps Now.
>Build and deploy apps & Web services for Linux with
>a free DVD software kit from IBM. Click Now!
>http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click
>_______________________________________________
>Linuxsampler-devel mailing list
>Lin...@li...
>https://lists.sourceforge.net/lists/listinfo/linuxsampler-devel
>
>  
>