Re: Frequency Inversion in MP3
- From: "coolup" <coolup012@xxxxxxxxxxxxx>
- Date: 31 Oct 2005 08:39:06 -0800
Jim Thomas wrote:
> coolup wrote:
> > hi All
> > In MP3 encoder algorithm, after the PQMF filtering, there is a step in
> > which the odd samples in odd subbands are multiplied by
> > -1. After this the MDCT is applied. Does any one here have any idea why
> > this is done.
> > I think it will invert the frequencies, that is lower
> > frequencies become higher and vice versa. But why is it done I don't
> > understand. Thanks for any replies.
> >
>
> In effect, multiplying odd samples by -1 is the same as (efficiently)
> multiplying the sequence by 1,-1,1,-1,...
>
> That is, muliplying by a cosine wave whose frequency is fs/4, where fs
> is the sample rate.
>
> In short, it's a mixer.
>
> --
> Jim Thomas Principal Applications Engineer Bittware, Inc
> jthomas@xxxxxxxxxxxx http://www.bittware.com (603) 226-0404 x536
> Today is the last day of your life so far.
Thanks for your reply Jin. I think the frequency of the
signal is fs/2, not fs/4.
But what I wanted was why is it done.
Why can't MDCT can be taken directly?
.
- References:
- Frequency Inversion in MP3
- From: coolup
- Frequency Inversion in MP3
- Prev by Date: Re: Fourier Transform Question for Pulsed CW and 802.11b DSSS
- Next by Date: Re: Motion Detector Unit For Loud Objects
- Previous by thread: Frequency Inversion in MP3
- Next by thread: Re: Reed-Solomon example in 802.16 standard
- Index(es):