attenuation - please explain

diji1 · 11th December 2001, 13:20

Hello,

I perform sample rate conversion on
my dvd soundtracks using ssrc after
decoding the ac3 using ssrc (48kHz -> 44.1 kHz).

Theres an option for attenuation - can
someone tell me what that is, and what it does
to the sound??

If ppl use it, how about some settings ??

Cheers!

LigH · 13th December 2001, 15:00

A sample rate conversion needs to calculate new sample points between existing ones. Due to the harmonic nature of audio, most HQ sample rate converters do not use a linear interpolation between two sample points, but instead an interpolation of a higher polynomial order or a sinusoidal synthesis. Unfortunately, these technologies can lead to clipping.

Let's think about a simplified example - imagine four sample points: The second and the third are very near the maximum, the first and forth a bit lower. If the converter calculated that it would need a new sample point between the second and third, it will calculate a curve which goes nicely through these four points; the top of this curve may now leave the allowed value range and would require this new sample point to be moved down, back into the valid range. This will lead into distortions (if they are audible, is another question).

To avoid this, you can have a first run where the application remembers how far the worst value is leaving the range, then the whole audio is attenuated (multiplied down) so that after the second pass the worst value just touches the maximum. - If you don't have the time for 2 passes, you may enter a custom attenuation value and hope that neither one value is still clipping nor the whole audio becomes too weak...

Shinigami-Sama · 24th February 2005, 08:31

thats what I figured interpolation was
could you possilbe pm a link or something with more info?
I;ve tried googling but as we all know to many results to find anything usefull..
I;ve been looking for a few days now
audio stuff like intregs me
well almost everything here does
but i;ve been looking for this for a while
might help me understand hte sample rate:qual things too..
~edit
wow necro post >.>

11th December 2001, 13:20	#1 \| Link
diji1 LOL-a-matic Join Date: Dec 2001 Location: The largest country in Region 4. Posts: 470	attenuation - please explain Hello, I perform sample rate conversion on my dvd soundtracks using ssrc after decoding the ac3 using ssrc (48kHz -> 44.1 kHz). Theres an option for attenuation - can someone tell me what that is, and what it does to the sound?? If ppl use it, how about some settings ?? Cheers! __________________ "The very rich are different from you and me" - F. Scott Fitzgerald "Yeah, they have more money" - Ernest Hemingway "What are you doing this weekend, Ania" - me ( damn she's beautiful ... ). "It is dangerous to be right in matters on which the established authorities are wrong." - Voltaire "movmasty is and was always right!"

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

13th December 2001, 15:00	#2 \| Link
LigH German doom9/Gleitz SuMo Join Date: Oct 2001 Location: Germany, Gardelegen Posts: 1,464	A sample rate conversion needs to calculate new sample points between existing ones. Due to the harmonic nature of audio, most HQ sample rate converters do not use a linear interpolation between two sample points, but instead an interpolation of a higher polynomial order or a sinusoidal synthesis. Unfortunately, these technologies can lead to clipping. Let's think about a simplified example - imagine four sample points: The second and the third are very near the maximum, the first and forth a bit lower. If the converter calculated that it would need a new sample point between the second and third, it will calculate a curve which goes nicely through these four points; the top of this curve may now leave the allowed value range and would require this new sample point to be moved down, back into the valid range. This will lead into distortions (if they are audible, is another question). To avoid this, you can have a first run where the application remembers how far the worst value is leaving the range, then the whole audio is attenuated (multiplied down) so that after the second pass the worst value just touches the maximum. - If you don't have the time for 2 passes, you may enter a custom attenuation value and hope that neither one value is still clipping nor the whole audio becomes too weak...