19 Feb
2004
19 Feb
'04
1:25 a.m.
I'm searching for a way to resample audio while preserving the highest
quality possible. One possible target would be the 96 --> 44.1
conversion, for obvious reasons, but other ratios are not excluded.
I'm speaking mostly from an "audiophile" perspective; raw numbers are
good ("while using this tool, you preserve such-and-such signal to noise
ratio and such-and-such bandwidth...") but i'd also like to hear about
musical tests, if anyone performed anything like that.
What would be the Linux tools that are supposed to provide the
highest-quality resampling (according to the above definition of
"quality")?
libsamplerate?
Something else?
--
Florin Andrei
http://florin.myip.org/
19 Feb
19 Feb
3:37 a.m.
On 18 Feb 2004 21:20:52 -0800
Florin Andrei <florin(a)andrei.myip.org> wrote:
I'm searching for a way to resample audio while
preserving the highest
quality possible. One possible target would be the 96 --> 44.1
conversion, for obvious reasons, but other ratios are not excluded.
I'm speaking mostly from an "audiophile" perspective; raw numbers are
good ("while using this tool, you preserve such-and-such signal to noise
ratio and such-and-such bandwidth...") but i'd also like to hear about
musical tests, if anyone performed anything like that.
What would be the Linux tools that are supposed to provide the
highest-quality resampling (according to the above definition of
"quality")?
libsamplerate?
There are two aspects to resampling:
a) maximizing signal to noise
b) maximizing passband width
If you can find anything that beats libsamplerate in both of these areas,
I want to know about it.
Erik (author of libsamplerate :-))
--
+-----------------------------------------------------------+
Erik de Castro Lopo nospam(a)mega-nerd.com (Yes it's valid)
+-----------------------------------------------------------+
Percussive Maintenance: The fine art of whacking the c**p out
of an electronic device to get it to work again.
6:06 a.m.
On Thu, 19 Feb 2004 18:09:25 +1100
Erik de Castro Lopo <erikd-lad(a)mega-nerd.com> wrote:
On 18 Feb 2004 21:20:52 -0800
Florin Andrei <florin(a)andrei.myip.org> wrote:
> libsamplerate?
I should also mention that there is an example program in the
libsamplerate source code tree called sndfile-resample. YOu
should be able to guess its function.
Erik
--
+-----------------------------------------------------------+
Erik de Castro Lopo nospam(a)mega-nerd.com (Yes it's valid)
+-----------------------------------------------------------+
Laws of Thermodynamics:
1) You cannot win.
2) You cannot break even.
3) You cannot get out of the game.
2:04 p.m.
On Wed, 2004-02-18 at 23:09, Erik de Castro Lopo wrote:
There are two aspects to resampling:
a) maximizing signal to noise
b) maximizing passband width
If you can find anything that beats libsamplerate in
both of these areas,
I want to know about it.
Indeed, libsamplerate looks very good from this perspective, that's why
i was investigating it.
But i'm also interested to know if anyone did a "live" test with various
resampling tools, and listen to high-quality records resampled by those
tools, ideally in a blind test.
I was kinda planning to do that myself some time in the (not too near)
future, but other people's input is appreciated. Especially since such a
test is difficult to do properly: use the right equipment (studio
monitors, refference headphones, etc.) in the right conditions, find
real high-q music records that feature details which could be
sampling-sensitive (usually bright sounds with fast attacks), take the
time to repeat the blind test enough times to make it significant, and
so forth.
But yes, lacking this kind of input, and having to make a quick decision
based on numbers only, i'd probably go with libsamplerate.
--
Florin Andrei
http://florin.myip.org/
8:04 p.m.
On Wed, 2004-02-18 at 21:20, Florin Andrei wrote:
I'm searching for a way to resample audio while
preserving the highest
quality possible. One possible target would be the 96 --> 44.1
conversion, for obvious reasons, but other ratios are not excluded.
I'm speaking mostly from an "audiophile" perspective; raw numbers are
good ("while using this tool, you preserve such-and-such signal to noise
ratio and such-and-such bandwidth...") but i'd also like to hear about
musical tests, if anyone performed anything like that.
For the underlying theory and the "resample" program see also:
http://www-ccrma.stanford.edu/~jos/resample/
-- Fernando
7585
days inactive
7585
days old
linux-audio-user@lists.linuxaudio.org
4 comments
3 participants
participants (3)
-
Erik de Castro Lopo -
Fernando Pablo Lopez-Lezcano -
Florin Andrei