<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=us-ascii">
<meta name="Generator" content="Microsoft Word 14 (filtered medium)">
<style><!--
/* Font Definitions */
@font-face
{font-family:"Cambria Math";
panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
{font-family:Calibri;
panose-1:2 15 5 2 2 2 4 3 2 4;}
@font-face
{font-family:Tahoma;
panose-1:2 11 6 4 3 5 4 4 2 4;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0in;
margin-bottom:.0001pt;
font-size:12.0pt;
font-family:"Times New Roman","serif";}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:blue;
text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
{mso-style-priority:99;
color:purple;
text-decoration:underline;}
span.EmailStyle17
{mso-style-type:personal-reply;
font-family:"Calibri","sans-serif";
color:#1F497D;}
.MsoChpDefault
{mso-style-type:export-only;
font-family:"Calibri","sans-serif";}
@page WordSection1
{size:8.5in 11.0in;
margin:1.0in 1.0in 1.0in 1.0in;}
div.WordSection1
{page:WordSection1;}
--></style><!--[if gte mso 9]><xml>
<o:shapedefaults v:ext="edit" spidmax="1026" />
</xml><![endif]--><!--[if gte mso 9]><xml>
<o:shapelayout v:ext="edit">
<o:idmap v:ext="edit" data="1" />
</o:shapelayout></xml><![endif]-->
</head>
<body lang="EN-US" link="blue" vlink="purple">
<div class="WordSection1">
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D">Hey Kas,<o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D">There’s an insanely easy trick to avoid denormalization. Not sure if you’re aware of it.
<o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D">The trick: just add a very small bias to the input signal. Something in the order of 1E-10 or 1E-11 works well: many dB below audible level; and may orders
of magnitude away from denormalization range (- 1E-35 to +1E-35, roughly). <o:p>
</o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D">Why does it work? Denorms almost always happen as the output of a filter (typically a IIR filter) settles toward zero. The chance of actually hitting a denorm
value under any other circumstance are so vanishingly rare its not even worth considering. About 1 in 1E20 or so for typical small audio signal values. So if you add a tiny DC bias on the input signal, then the ouput filter should settle toward an even tinier
DC bias on the output signal, and that tiny bias will be more than enough to keep the signal far away from denorm range.
<o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D">As a matter of reasonable practice, you should us different constant randomly-selected small positive or negative values every time you choose to add an anti-denormalization
bias, in order to avoid accidentally cancelling out the bias in successive stages.<o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D">The one counterexample to watch out for: IIR filters that have a perfect notch at 0Hz (many/most(?) high pass filters). In that case you want to add the denormalization
bias to one of output delay taps. <o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D"><o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497D"><o:p> </o:p></span></p>
<p class="MsoNormal"><b><span style="font-size:10.0pt;font-family:"Tahoma","sans-serif"">From:</span></b><span style="font-size:10.0pt;font-family:"Tahoma","sans-serif""> chuck-users-bounces@lists.cs.princeton.edu [mailto:chuck-users-bounces@lists.cs.princeton.edu]
<b>On Behalf Of </b>Kassen<br>
<b>Sent:</b> Monday, February 28, 2011 9:46 AM<br>
<b>To:</b> ChucK Users Mailing List<br>
<b>Subject:</b> Re: [chuck-users] clearing a delay line<o:p></o:p></span></p>
<p class="MsoNormal"><o:p> </o:p></p>
<p class="MsoNormal" style="margin-bottom:12.0pt"><o:p> </o:p></p>
<div>
<p class="MsoNormal">On 28 February 2011 15:35, Casper Schipper <<a href="mailto:casper.schipper@gmail.com">casper.schipper@gmail.com</a>> wrote:<o:p></o:p></p>
<p class="MsoNormal" style="margin-bottom:12.0pt">I wondered if there is any method to clear the internal buffer of a Delay ugen (since I did find such a function in the STK-kit).<o:p></o:p></p>
<div>
<p class="MsoNormal"><o:p> </o:p></p>
</div>
<div>
<p class="MsoNormal">Hey Casper.<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"><o:p> </o:p></p>
</div>
<div>
<p class="MsoNormal">Try setting the max length again. I believe that should flush the buffer as the memory is cleared and re-initialised. To do it without flushing you could use LiSa as a delay. LiSa has a explicit clear function. Theoretically speaking that
might prevent memory fragmentation and should have a lower performance cost (and you can ramp down before clearing to avoid clicks). In practice I seem to remember that clearing LiSa can lead to some performance costs too, though that might be related to my
ongoing struggle with hard to detect floating-point de-normalisation issues.<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal"><o:p> </o:p></p>
</div>
<div>
<p class="MsoNormal">Hope that helps,<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal">Kas.<o:p></o:p></p>
</div>
</div>
</div>
</body>
</html>