Re: a question about phase vocoder (Kun She )


Subject: Re: a question about phase vocoder
From:    Kun She  <intel_ghost(at)SOHU.COM>
Date:    Thu, 29 Dec 2005 11:34:22 +0800

------=_Part_1284_32179480.1135827262448 Content-Type: multipart/alternative; boundary="----=_Part_1283_1345529.1135827262447" ------=_Part_1283_1345529.1135827262447 Content-Type: text/plain; charset="GB2312" Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by torrent.cc.mcgill.ca id jBT3YjhU021819 Hi,Jim and everyone!<p><p>Please pardon me. Maybe I didn't make myself un= derstood.<p><p>If explaining STFT as linear filtering, then its hop size<= p>must be set according to the type of analysis window and <p>widow lengt= h so that the original signal can be restored from the time<p>downsampled= STFT. It is exactly right!<p>My question is:<p>The outputs of phase voco= der are time-variable ampltitude<p>(or say, instantaneous amplitude) and = instantaneous frequency.<p>In the discrete-time case, instantaneous frequ= ency have to be<p>computed as difference of phase or as given by <p>(J. L= . Flanagan and R. M. Golden, =A1=B0Phase vocoder,=A1=B1). <p>In the first= case, phase must be extended beyond <p>the principal value interval [-pi= ,pi]. If the hop size<p>is 1, this job is easy. If not, how to do then?<p= >In the latter case, differential of STFT's real part and<p>imaginary par= t have to be estimated by discrete-time<p>filtering of them, for example,= the first order forward or backward<p>difference . If the hop size is not 1, differential can still be replaced<p>by diff= erence? The real part and imaginary part of STFT are<p>band-limited in w= hat extent? How to set a reasonable hop size?<p> <p><p>Best<p> Regar= ds!<p><p>Superk (Kun She) <p>intel_ghost(at)sohu.com<p>2005-1= 2-29 ------=_Part_1283_1345529.1135827262447 Content-Type: text/html Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by torrent.cc.mcgill.ca id jBT3YjhU021819 <html> <head> <meta http-equiv=3D"Content-Type" content=3D"text/html; charset=3DGB2312"= ></head> <body> <!--SOHUMAIL_HTML_HEAD_END-->Hi,Jim and everyone!<p><p>Please pardon me. = Maybe I didn't make myself understood.<p><p>If explainiing STFT as linear= filtering, then its hop size<p>must be set according to the type of anal= ysis window and <p>widow length so that the original signal can be restor= ed from the time<p>downsampled STFT. It is exactly right!<p>My question i= s:<p>The outputs of phase vocoder are time-variable ampltitude<p>(or say,= instantaneous amplitude) and instantaneous frequency.<p>In the discrete-= time case, instantaneous frequency have to be<p>computed as difference of= phase or as given by <p>(J. L. Flanagan and R. M. Golden, =A1=B0Phase vo= coder,=A1=B1). <p>In the first case, phase must be extended beyond <p>the= principal value interval [-pi,pi]. If the hop size<p>is 1, this job is e= asy. If not, how to do then?<p>In the latter case, differential of STFT's= real part and<p>imaginary part have to be estimated by discrete-time<p>f= iltering of them, for example, the first order for ward or backward<p>difference. If the hop size is not 1, differential ca= n still be replaced<p>by difference? The real part and imaginary part of= STFT are<p>band-limited in what extent? How to set a reasonable hop siz= e?<p> <p><p>Best<p> Regards!<p><p>Superk (Kun She) <p>= intel_ghost(at)sohu.com<p>2005-12-29<hr size=3D1><style type=3D"text/css"> <!-- a {font-size:12px;color:black} a:hover {font-size:12px;text-decoration:underline} .b{font-family:Arial; font-size:80%; margin-bottom:2px;color:#0000CC} .pp{font-family:Arial;font-size:10px} --> </style> <table border=3D0> <!--tr><td height=3D18><a href=3Dhttp://goto.sohu.com/goto.php3?code=3Dma= iladt-ta1 target=3D_blank>=CB=D1=BA=FC=D0=A1=CE=C0=CA=BF=A1=AA=A1=AA=C8=AB= =D0=C2=B7=B4=C0=AC=BB=F8=CF=B5=CD=B3</a> <br> </td></tr--> <tr><td height=3D18><a href=3Dhttp://goto.sohu.com/goto.php3?code=3Dmaila= dt-ta1 target=3D_blank><font color=3Dred>=D6=BB=D2=AA=CC=E1=D2=E2=BC=FB=BE= =CD=C4=DC=B5=C3=B4=F3=BD=B1=A3=A1</font></a> </td></tr> <br> <tr><td height=3D18><a href=3Dhttp://goto.sohu.com/goto.php3?code=3Dmaila= dt-ta target=3D_blank>SOHU.COM=CD=AC=C3=FB=C9=FD=BC=B6VIP=A3=AC=CE=D2=B5=C4= =C3=FB=D7=D6=CE=D2=D7=F6=D6=F7=A3=A1</a> </td></tr> </table><!--SOHUMAIL_HTML_TAIL_END--></body> </html> ------=_Part_1283_1345529.1135827262447-- ------=_Part_1284_32179480.1135827262448--


This message came from the mail archive
http://www.auditory.org/postings/2005/
maintained by:
DAn Ellis <dpwe@ee.columbia.edu>
Electrical Engineering Dept., Columbia University