Re: [AD] sound frequency?

[Mike Fahlbusch <mcf@xxxxxxxxxx>]

At 07:54 AM 7/05/2004, you wrote:

> When testing the digmid patch before.. I noticed something strange:
> Allegro's ALSA driver would pitch my .mids slightly higher than the OSS
> driver. I found out, the difference is due to the default frequency:
> ALSA uses 44100, OSS 45454.
>
> I have no idea how sound works.. but still, I'm wondering now. Why do
> different frequency values change the pitch of a midi, and which value
> should be used as default? Different values for OSS/ALSA as right now
> probably is not the right thing.
>
> Anyone can shed light on this?

44100 Hz is the correct frequency.


> Ben Davis wrote:
> > The figure 45454 comes from the early Sound Blasters.  It wasn't possible to
> > specify a precise frequency
>
> Just so you know, the SB16 was immune to this. I couldn't test it on any 
> other Sound Blasters, but when I had an SB16, I modified the DOS SB16 
> driver, a long time ago durring me toying with code phase, so it used the 
> exact frequency requested. I tried 44100, 22050, 16000, 11025, and even 
> 8000, and they all sounded right on pitch next to their off-value 
> counterparts (plus on my old 486 those few Hz really heklped performace 
> with DIGMID :P). I'm not exactly sure what I did though, so I'm not sure I 
> can do the same changes again.

(To Elias Pschernig)
On many soundcards, 44100 Hz is the base or natural frequency.  The reason 
is that the analog electronics in the soundcard is designed to give good 
sound quality and accurate reproduction at 44100 Hz.  When a different 
frequency is programmed, they reprocess the 44100 Hz digital sample to 
give, say, 22050 Hz.  This means that oddball frequencies can't be 
programmed into the soundcard because they can't be divided into 44100 
Hz.  So 11025 Hz might be acceptable, 11000 Hz isn't.
In my experience older soundcards like the sb16 could be programmed at 
almost any frequency as long as it was a multiple of 128 (I think it was 
128).  Modern soundcards however reprocess the sound card using a DSP and 
usually can't be programmed to use odd frequencies.
When a sound recording program was taken from a Pentium sb16 and ported to 
a new Athlon XP system, the new crappy onboard sound chip wouldn't accept a 
frequency of 16384 Hz.  This freq was selected because it was convenient 
for doing an FFT (FFT sample sizes must be a multiple of 2).  This meant 
the recording freq had to be changed to 22050 Hz before the program would work.

Playing back a sound at a higher speed makes the pitch of the sound go 
up.  Funnily enough, playing back a sound at a lower speed makes the pitch 
of the sound go down.  The allegro\tests\akaitest.exe program uses this 
technique to play the same sound at different pitches.  (On my computer 
under W98SE I have to run allegro\setup\setup.exe, copy the saved 
allegro.cfg to allegro\tests, change digi_card = AXA  midi_card = -1.)

(Science lesson time) Doppler shift sound is actually done by an algorithm 
which increases or decreases the playback speed of a sound.  It's the same 
sound, but if it is played back faster, it sounds like the object is coming 
nearer.  Decreasing the playback speed makes the objects sound like it is 
moving away.  If the objects if moving faster, the playback speed is 
increased even more.  An example of a game with doppler shift is Battlezone 
II, approaching tanks have their engine sound doppler shifted for more 
realistic sound effects.  Some games even go as far to make the doppler 
shifted sound move from the speaker to the other.  Soundcards after EAX 
standard usually seem to have doppler shifting included.

If more info is needed, google for  cd sampling rate  and it will give lots 
of explanations about why various sound sampling frequencies are chosen.

Sig:
When people see a cow, some want milk and others want steak.  One feeds a 
person for years, the other only a few weeks.