Re: [LAU] Popping sound at the start of every note on Bristol synths.

Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]
To: <lsd@...>, <linux-audio-user@...>
Date: Monday, March 12, 2012 - 9:08 am

--_157f0c97-0c25-409a-98ac-7f03947b860b_
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

> Date: Mon=2C 12 Mar 2012 09:11:46 +1100

nths.
ry
ried ALSA
e=20
nd=20

I had a listen through the sample you posted and agree that it is an issue =
withthe attack time=2C it was always very aggressive but there was also a r=
ecent changefrom a default exponential to default linear attack which proba=
bly does not helpwith what you are hearing. You can change back to exponent=
ial from the configstage however there is _another_ change that will go int=
o the next release whichis probably the correct fix.
The change is to make the envelope 'time constant'. This is actually not th=
e casewith the current code so if you increase the sampling rate you also i=
ncrease thefastest attack rate. The fix places the attack at 500us then giv=
es you a rate which is independent of the sampling rate. I can make this a =
config time option if peopleare interested=2C it could be a runtime option =
which I think that is overkill but 500us is still pretty aggressive.
The maximum attack duration does not change=2C that is also a config option=
which defaults to 10 seconds however having the parameters being a functio=
n of the sampling rate was bad coding.
FYI some background on the changes: I have been doing work on 'low end'syst=
ems where I default to half rate sample=2C 22KHz and 24KHz to reduce the CP=
Ufootprint and hence extend battery life. At the same time I have been work=
ing onhigh end systems where we have been reviewing the improved sound qual=
ity with96KHz and 192KHz and these really exacerbated the problem. There ar=
e other interesting results from the developments. The main filters are Huo=
vilainen's whichare 2x resampling=2C this was required since the algorithm =
loses frequency accuracyas it approaches Nyquist. At the higher rates I rem=
oved the resampling since the cutofffrequency is now nowhere near Nyquist a=
nd the result is that=2C for the filters at least=2Cthe change from 48 to 9=
6KHz is almost free and you can still 'play' the filter at selfoscillation.
Let me review the code differences and perhaps post the new envelope.c on t=
heSourceForge forum. That might not be possible: the envelopes actually dri=
ve thesignalling to the engine regarding when to deactivate voices in polyp=
honic voiceassignments and I don't want any more sticky or lost notes where=
this signallingmight have changed.
Regards=2C nick. =

--_157f0c97-0c25-409a-98ac-7f03947b860b_
Content-Type: text/html; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

&gt=3B Date: Mon=2C 12 Mar 2012 09:11:46 +1100&gt=3B From: lsd@woo=
tangent.net&gt=3B To: linux-audio-user@lists.linuxaudio.org&gt=3B S=
ubject: Re: [LAU] Popping sound at the start of every note on Bristol synth=
s.&gt=3B &gt=3B On 12/03/12 9:03 AM=2C Rafael Vega wrote:&gt=3B=
&gt=3B However=2C I've found that there is a popping sound at the start of=
every&gt=3B &gt=3B note. This is more apparent in some models/presets =
than others. I'm&gt=3B &gt=3B pretty sure my jack configuration is fine=
(has been working fine with&gt=3B &gt=3B other synths=2C ardour=2C pur=
edata=2C qtractor=2C etc.) and I've also tried ALSA&gt=3B &gt=3B and tw=
o different sound cards (laptop internal and firewire). I also&gt=3B &g=
t=3B tried outputting Bristol straight to system output=2C passing through<=
br>&gt=3B &gt=3B It may just be that the attack on the amplitude envelo=
pe is too fast. &gt=3B I'm not sure if Bristol has this problem=2C but =
on many synths=2C with the &gt=3B attack time set to its minimum=2C the=
envelope opens instantaneously=2C and &gt=3B that can cause a clicking=
sound. Try increasing the attack time and see &gt=3B if that helps.&gt=3B &gt=3B ThanksI had a listen through the samp=
le you posted and agree that it is an issue withthe attack time=
=2C it was always very aggressive but there was also a recent change<=
div>from a default exponential to default linear attack which probably does=
not helpwith what you are hearing.&nbsp=3BYou can change back to exponential from the configstage however there is _another_ change that will go into the next =
release whichis probably the=
correct fix.The change is to make the=
envelope 'time constant'. This is actually not the case<=
span style=3D"font-size: 10pt=3B ">with the current code so if you increase=
the sampling rate you also increase thefastest attack rate. The fix places the attack at 500us =
then gives you a rate which&nbsp=3Bis independent of the sampling rate. I can make this a config=
time option if peopleare interested=2C it could be a runtime option which I think that is overk=
ill but 500us&nbsp=3B=
is&nbsp=3Bstill pretty aggressiv=
e.The maximum attack duration does not=
change=2C that is also&nbsp=3Ba=
config&nbsp=3Boption which&nbsp=
=3Bdefaults to 10 sec=
onds however having the parameters being a function of the&nbsp=3Bsampling rate was bad coding.<=
/span>=
FYI some background on the changes: I have been doing work on 'low end'systems where I default to half rate sample=2C 22KHz and 24KHz to re=
duce the CPUfootprint and hence extend battery life. At the same=
time I have been working onhigh end systems where we have been =
reviewing the improved sound quality with96KHz and 192KHz and th=
ese really exacerbated the problem. There are other&nbsp=3Binter=
esting results from the developments. The main filters are Huovilainen's wh=
ichare 2x resampling=2C this was required since the algorithm lo=
ses frequency accuracyas it approaches&nbsp=3BNyquist. At the hi=
gher rates I removed the resampling since the cutofffrequency is=
now nowhere near&nbsp=3BNyquist&nbsp=3Band the result is that=2C for the f=
ilters at least=2Cthe change from 48 to 96KHz is almost free and=
you can still 'play' the filter at selfoscillation.<=
br>Let me review the code differences and perhaps post the new e=
nvelope.c on theSourceForge forum. That might not be possible: t=
he envelopes actually drive thesignalling&nbsp=3Bto the engine r=
egarding when to deactivate voices in polyphonic voiceassignment=
s and I don't want any more sticky or lost notes where this signallingmight have changed.Regards=2C nick. =

=

--_157f0c97-0c25-409a-98ac-7f03947b860b_--

Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]

Messages in current thread:
Re: [LAU] Popping sound at the start of every note on Bristo..., Nick Copeland, (Mon Mar 12, 9:08 am)