Subject: envelope extraction From: David Smith <smithd@xxxxxxxx> Date: Fri, 27 Feb 2009 00:18:25 -0500 List-Archive:<http://lists.mcgill.ca/scripts/wa.exe?LIST=AUDITORY>This is a multi-part message in MIME format. --_----------=_1235711905182453 Content-Disposition: inline Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="iso-8859-1" Re: Method for generating an envelope around an audio signal: Low pass filtering was pretty good back in the day. A more modern DSP method would be to connect consecutive positive peaks in the audio signal with straight lines or connect peak groups with bezier curves. Then perhaps apply low pass filter at 15 - 20kHz. The advantage of this method is that it preserves indication of fast transients which we hear but are removed by 50-400 Hz low pass filters.=20 I suppose half wave rectification before finding consecutive peaks will yield a slightly more accurate envelope. It would also make sense to use RMS rather than peak values. I don't know of any off-the-shelf program to do this. Dave Smith roughlight.com Buy my music, I'm unemployed! --=20 Be Yourself @xxxxxxxx mail.com! Choose From 200+ Email Addresses Get a Free Account at www.mail.com --_----------=_1235711905182453 Content-Disposition: inline Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset="iso-8859-1" <br><br>Re: Method for generating an envelope around an audio signal:<br><b= r>Low pass filtering was pretty good back in the day. A more <br>mode= rn DSP method would be to connect consecutive positive <br>peaks in the aud= io signal with straight lines or connect peak <br>groups with bezier curves= . Then perhaps apply low pass filter <br>at 15 - 20kHz.<br> <br>The a= dvantage of this method is that it preserves indication of<br>fast transien= ts which we hear but are removed by 50-400 Hz <br>low pass filters. <= br><br>I suppose half wave rectification before finding consecutive <br>pea= ks will yield a slightly more accurate envelope. It would <br>also ma= ke sense to use RMS rather than peak values.<br><br>I don't know of any off= -the-shelf program to do this.<br><br>Dave Smith<br>roughlight.com<br>Buy m= y music, I'm unemployed!<br><BR> --=20 <div> Be Yourself @xxxxxxxx mail.com!<br> Choose From 200+ Email Addresses<br> Get a <b>Free</b> Account at <a href=3D"http://www.mail.com/Product.aspx" t= arget=3D"_blank">www.mail.com</a>!</div> --_----------=_1235711905182453--