[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [fluid-dev] SoundFont exclusiveClass

From: Dan Eble
Subject: Re: [fluid-dev] SoundFont exclusiveClass
Date: Mon, 12 Sep 2016 13:10:34 -0400

On Sep 12, 2016, at 09:32 , S. Christian Collins <address@hidden> wrote:
> On 09/12/2016 08:13 AM, Dan Eble wrote:
>>>> Say I have an instrument with a release time of 2 seconds.  Fluidsynth 
>>>> sounds as if it is playing the sound at constant volume for 2 seconds 
>>>> after the end of the note, then stopping abruptly.  According to the 
>>>> SoundFont spec, the volume should instead decrease linearly to zero.
>>> I have been working with FluidSynth for years and have never encountered 
>>> this issue. Do you have a SoundFont that exhibits this problem that you 
>>> could share?
>> Here’s a simplified version.  The instrument I’m having trouble with is 
>> supposed to give the impression of the sympathetic resonance of a cello 
>> string when another string is bowed.  It’s called “Vc, Bowed Symp. 4” in the 
>> attached file.
> The problem isn't the envelope, it's the excessive amount of filter Q you are 
> using on that layer. What appears to be happening is that the 96 dB of 
> resonance amplifies the resonant frequency to a super-high level, which seems 
> to be getting capped by a built-in limiter. Because the resonant level is so 
> high, the limited audio level sounds flat. Since a voice at its quietest 
> envelope point will receive 96 dB of attenuation according to the SF spec, 
> the voice is essentially 100% amplified until it is cut off (which is why you 
> hear a click).
> So, if you reduce the filter Q level, you will begin to hear the proper 
> release phase.

Is that consistent with the diagram in the spec (version 2.04, section 9.1.8)?  
The diagram shows the volume envelope applied to the filter output.  I don’t 
understand why the volume shouldn’t therefore decrease linearly during the 
release phase (maybe because I’m new at this).  Even if it begins very at some 
high value, shouldn’t still diminish at a constant rate?


reply via email to

[Prev in Thread] Current Thread [Next in Thread]