Re: [AD] [SCM] Allegro branch 5.1 updated. 5.1.3-21-g1a7e633

[Peter Wang <novalazy@xxxxxxxxxx>]

On Mon, 3 Sep 2012 19:47:00 +0200, Evert Glebbeek <eglebbk@xxxxxxxxxx> wrote:
> On 3 Sep 2012, at 14:52 , Peter Wang wrote:
> > I think I implemented it correctly, but it sounds really bad (worse than
> > linear interpolation).  Frequency plot attached.  But perhaps you can
> > have a look over the patch and see if I got it completely wrong.
> 
> The code looks correct to me, but I'm not sure that I interpret correctly what value0-3 are or their relation to the variable t. I'd have to study the code a bit closer to understand it properly.

value0-3 are the values at sample positions i-1 through i+2,
where we are trying to interpolate the value at i+t.
t ranges from 0.0 to 1.0.

> The frequency plot is a bit hard to interpret for me, have you looked at the waveform?

I generated a 490 Hz sine wave or thereabouts (not exactly 500 Hz)
in a stream of frequency 8000 Hz, which is then resampled to 44100 Hz.
The big peak is the sine wave.  The rest are unwanted artefacts.

I did look at the waveform yesterday, but not closely enough.
Attached a screenshot when I zoomed in.

Peter

Attachment: ss2.png
Description: PNG image