Re: [linux-audio-dev] Resampling

New Message Reply About this list Date view Thread view Subject view Author view Other groups

Subject: Re: [linux-audio-dev] Resampling
From: Vincent Touquet (vincent_AT_ulyssis.org)
Date: Sat Aug 04 2001 - 00:20:43 EEST


(cut)
>If it's integer only, then yes it will be really easy to implement and the
>quality will be good.
>For fractional values I have no clue how to do this (eg: "resample the
>frequency bands ? ))
(cut)

For fractional values, you generally do the upsampling first and then the downsampling (to avoid 'losing' data). The filter after the upsampling and before the downsampling can be merged into one filter in between. The problem is here that you do computations which can be avoided: you should do your computations at the lowest rate possible (you upsample and you apply a filter (: too fast) and afterwards you throw some samples away by downsampling). Both type of redundancies can be avoided by using the proper multiphase decomposition of the system function of the filter. Using the theorem of Euclid stating that the gcd of two relative prime values is one; 1 = p.m + q.n (p,q E Z), you can combine both decompositions to effectively get the computations at the lowest rate possible. *No* computations are saved, but they are to be done on a lower frequency, which is nice :)Note: m and n are the up and downsampling numbers.

If you want more details, shout (this is from my dsp course part I, which I had this year :)

Kind Regards
vincent

PS: sorry if I omitted to much and if I'm not clear i typed it out rather fast


New Message Reply About this list Date view Thread view Subject view Author view Other groups

This archive was generated by hypermail 2b28 : Sat Aug 04 2001 - 00:21:42 EEST