Re: [Linuxsampler-devel] release trigger bug

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

Hi,

I have know examined and hopefully corrected this old issue:

On 2004-12-30 19:20, Christian Schoenebeck wrote:
> Es geschah am Donnerstag 30 Dezember 2004 19:10 als Garett Shulman schrieb:
> 
>>Hey Mark, I don't think that the time between noteon and noteoff events
>>makes a difference. In fact, the noteon event should be irrelevent
>>(these seem to be playing pretty well). The noteoff event triggers a
>>completely different sample than an equivelent noteon event on all of
> 
> 
> Correct
> 
> 
>>the patches I have. Do you happen to have Scarbee RSP 73? This gig uses
>>noteoff triggered samples to simulate the sound of the damper striking
>>the tine as the tine is silenced when the key is released. This should
>>be quantifiable by comparing the amplitude of recordings you make from
>>ls and gs in an audio editor (like snd or soundforge). I don't think
>>you need to bother generating noteon events at all. Is this making
>>sense? -Garett
> 
> 
> Actually this is a good question. In regards of LS you are right, when the 
> note-off arrives we play a completely independent sample and determine the 
> right sample to play (together with its articulation information) by its key 
> number, the positive release trigger dimension (release trigger dimension is 
> a binary value, false = sample for note on, true = sample for note off) and 
> its release trigger velocity (actually it depends on couple more things, but 
> these can be ignored for our problem). But I'm not sure if this is correct. 
> Maybe the original GSt uses the original note on velocity instead of the note 
> off velocity? At least that would explain why LS and GSt behave differently 
> regarding release triggered samples.
> 
> Mark, could you please determine this? I mean if the release triggered sample 
> on Gst is selected by its releease velocity or by its previous note on 
> velocity?

I did some measurements which show that the volume of the release 
triggered sample is derived from three values: the previous note-on 
velocity, the time from note-on to note-off (the note length) and the 
value of the "Release trigger decay" parameter in the gig file. The MIDI 
note-off velocity seems to be ignored.

I have committed a fix for this.

/Andreas