Re: Generating sine waves with Winsound

[Al Bregman <al.bregman@xxxxxxxxx>]

Dear Marco,

I analyzed your signal and estimated that in the sample you sent me the
two tones were at 300 and 325 Hz, respectively and that the onsets and
offsets were 10 ms in duration.  During the onsets and offsets, I  found
weak energy at about 4100, 4700, 8400, and 9100 Hz.  These anomalies
would be too weak to hear in normal listening, and I didn't really hear
anything that I would call clicks.  Naturally with 10-ms rise times the
onsets won't be really smooth.

I then synthesized equivalent signals on my software and found no energy
at the frequencies that were present in your signal.  I conclude that
the problem was in your synthesis software,  I found the extra energy
only at the onsets and offsets, so the problem must be in the software
components that shape these events.

Best,

Al Bregman

. ----- Original Message -----
From: "Marco Kobbenbring" <marco.kobbenbring@gmx.net>
To: <p.kusmierek@NENCKI.GOV.PL>; <avc@pacific.net.ph>;
<pallier@lscp.ehess.fr>; <eliot@generation.net>;
<marta_andrea@libero.it>; <regis@lsi.usp.br>; <al.bregman@mcgill.ca>;
<joby@dsplab.ece.iisc.ernet.in>; <masta@UMICH.EDU>;
<stefan.muenkner@uni-tuebingen.de>;
<Eckard.Blumschein@E-Technik.Uni-Magdeburg.DE>
Sent: Monday, January 12, 2004 3:03 PM
Subject: Generating sine waves with Winsound


> Hello!
>
> Many thanks for your help with my problem of clicking sine waves.
> Fortunately, with the assistance, I was able to solve it.
> At first glance, the clicking was not "visible" in the wave form, but
it was
> in a
> spectrogram.
> At the begining and at the end of each tone, there were shadows in
high
> frequency
> regions.
> I just filtered everything above 2kHz, and the clicks disappeared.
> However, this does not remove the actual cause of the clicks, which
> obviously lies
> in the synthesis algorithm. Funny is that they coincide with the rise
and
> decay
> times of the tones.
> To this email I attached a short example which shows the shadows in
the
> spectrogram
> test_120104.wav. I would appreciate if you may have ideas about their
> origin.
>
> Best wishes!
>
> Marco Kobbenbring
>
> --
> -----------------------------------
> Marco Kobbenbring
> Steinstr. 8
>
> 24118 Kiel
>
> EMail: marco.kobbenbring@gmx.net
> Telefon: 0431/8885966
> -----------------------------------
>
> +++ GMX - die erste Adresse für Mail, Message, More +++
> Neu: Preissenkung für MMS und FreeMMS! http://www.gmx.net
>