<html><body><div style="color:#000; background-color:#fff; font-family:arial, helvetica, sans-serif;font-size:12pt">Dear Stefan,<br>In the paper "Improved Forward-Adaptive Prediction for MPEG-4 Audio Lossless Coding", a non-linear compander is applied to the parcor coefficients prior to quantization. This compander is designed in order to minimize quantization error, especially for magnitudes close to unity. <br>If you determine the typical distribution of magnitudes of the LPC coefficients, you could design a good non-linear compander in order to minimize the error due to LPC coefficients quantization. By doing this, the input signal would be better estimated and the output residue (which occupies more than 90 % of the output bitrate) would be smaller.<br>Regards,<br><br>Fernando A. Marengo Rodriguez<br>Acoustics and Electroacoustics Laboratory<br>Faculty of Sciences, Engineering and Surveying<br>National University of Rosario<br>Rosario,
Argentina<br>http://www.fceia.unr.edu.ar/acustica/codecdigital/integrantes.html#Fernando<br><br><br>Message: 1<br>Date: Tue, 19 Jul 2011 14:48:27 +0200<br>From: Stefan Westerfeld <<a href="mailto:stefan@space.twc.de">stefan@space.twc.de</a>><br>Subject: [Flac-dev] Using line spectral pairs for LPC quantization<br>To: <a href="mailto:flac-dev@xiph.org">flac-dev@xiph.org</a><br>Message-ID: <<a href="mailto:20110719124826.GB1153@space.twc.de">20110719124826.GB1153@space.twc.de</a>><br>Content-Type: text/plain; charset=us-ascii<br><br> Hi!<br><br>I've recently read the flac sources, and it appears to me that you're quatizing<br>the LPC coefficients directly. However, this might be not as efficient, because<br>LPC coefficients are very sensitive to quantization error. I don't know how<br>much space the LPC coefficients occupy in the resulting FLAC file - if its not<br>much, it probably doesn't matter.<br><br>But if its a significant amount
of data, it might be better to convert the LPC<br>coefficients to LSF/LSP coefficients, and store those quantized. This should<br>provide the same quality at a lower bit count, if I'm right. Here is a link<br>that describes LSF/LPC coefficients:<br><br><a href="http://en.wikipedia.org/wiki/Line_spectral_pairs" target="_blank">http://en.wikipedia.org/wiki/Line_spectral_pairs</a><br><br> Cu... Stefan<br>-- <br>Stefan Westerfeld, Hamburg/Germany, <a href="http://space.twc.de/%7Estefan" target="_blank">http://space.twc.de/~stefan</a><br><div style="font-family: times new roman, new york, times, serif; font-size: 12pt;"><br></div></div></body></html>