[ Thread Index |
Date Index
| More lists.liballeg.org/allegro-developers Archives
]
On Sat, 21 Apr 2012 11:22:18 +1000, Peter Wang <novalazy@xxxxxxxxxx> wrote:
> On Sat, 21 Apr 2012 02:49:20 +0200, Elias Pschernig <info@xxxxxxxxxx> wrote:
> > On Sat, 21 Apr 2012 10:34:25 +1000
> > Peter Wang <novalazy@xxxxxxxxxx> wrote:
> > >
> > > Hi Elias,
> > >
> > > ex_resample_test sounds horrible now. In the visualisation I can see
> > > many gaps in the waveform. Perhaps you're still working on it?
> > >
> >
> > Hm, which driver are you using? It's only tested with pulse audio,
> > guess I need to try with something else. My first guess would be that
> > something goes wrong now when there's small buffers (our pulseaudio
> > driver always uses a fixed 4096 samples buffer).
>
> Both ALSA or OpenAL.
>
> >
> > Can you try revisions 15551, 15552 and 15554?
>
> 15550 "ok"
> 15551 "ok"
> 15552 "ok"
> 15554 bad
>
> By "ok", I mean that it doesn't sound worse than the original test.
> The original test does not produce a pure tone either, but that's a
> separate bug.
Here's another test. In ex_synth.cpp reduce the stream frequency to
8000 or 11025:
#define STREAM_FREQUENCY (8000)
then compile and turn on the sine wave generator.
With the old resampling algorithm, there is a periodic crackle (bug)
but otherwise a clean tone. With the new algorithm, there is the tone,
the crackle, and a very obvious high pitched, "metallic" noise component.
Peter