Message-ID: <3ADD5896.D024A120@usa.net>
Date: Wed, 18 Apr 2001 11:04:22 +0200
From: "Alberto di Bene" <dibene@usa.net>
MIME-Version: 1.0
To: rsgb_lf_group@blacksheep.org
Subject: LF: Re: WOLF test results, soundcard trouble
References: <a7.d9bf85b.280e1ae1@aol.com>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

DL4YHF@aol.com wrote:
> 
[snip] 
> The precise sample rates of the soundcard, running at a nominal sample rate
> of 8000 samples per second, were **DIFFERENT** for TX (D/A conversion) and RX
> (A/D conversion)  !
[snip] 

Wolf and the group,
     in the Windows world a new standard is emerging, the AC97 standard for 
sound cards. How does that affect us ? With AC97, the hardware samples *always*
at the maximum sampling rate of 44.1 or maybe 48 kHz. 
When a program opens the sound subsystem specifying a different sampling rate, 
the driver does a downsampling of the 44.1 (48) kHz stream when opening for input, 
and an upsampling, probably with zero-padding, when opening for output.

I have verified this in practice. On my IBM Thinkpad, I am able to start
simultaneously several programs that use the sound card (sound chip in this case),
and they work correctly even if they specify different sampling rates !
At first I didn't believe to my eyes, but then I read about the AC97 standard.

The implications of this are that in the process of down/up sampling, some roundoff
errors can happen, and the actual sampling rate, as seen by the application,
can differ by a slight amount from what specified. For the normal use of the 
sound cards, i.e. games, this has no consequences. Not so for a more serious use.
So the advice is to stay away, whenever possible, from sound cards whose drivers
implement the AC97 standard, which, I'm afraid, is what Bill want us to use.  

73  Alberto  I2PHD