[LAU] Need to convert 24bit/48000 to 16bit/44100 stereo on linux

Fons Adriaensen fons at kokkinizita.net
Mon Jan 14 13:00:17 EST 2008

On Mon, Jan 14, 2008 at 06:32:14PM +0100, Kjetil S. Matheussen wrote:
> Fons Adriaensen:
> > I wrote libzita-resample mainly for high-quiality real-time SRC
> > of multichannel audio, and it's a compromise between 'numerical'
> > quality and CPU load. But so far nobody has been able to hear the
> > difference with a no-compromise algorithm in a series of listening
> > tests (all audio experts and musicians).
> Interesting. Do you know if anyone is able to hear the difference
> between a 48000hz sound downsampled to 44100hz and then upsample
> to 48000hz again, and the original 48000hz sound?
> Just wondering. :-)

There was a very interesting report about this in the JAES some
months ago. I don't have it here, but will post the reference
later (I should have it at home). 

The authors have conducted quite an extensive series of listening
tests to find out if anyone could hear the difference between the
'High Resolution' audio formats and the plain old CD. They did it
by recoding DVD-A and SACD recordings to CD standards, and let the
listeners compare with the original. Equipment and test procedures
were up to the highest standards.

The conclusion was IIRC rather clear: you can't hear the difference.

Next question then was why all these HR recordings sounded so
much better than the average CD recording of the same music -
most listeners and the also the authors of the report did agree
on that.

The answer suggested by the authors is quite sobering: because
these recordings have been made for a niche market of audiophiles,
and the sound engineers who made them were therefore not subject
to the usual pressure to produce a type of sound that record
companies think sells best (reduced dynamic range, a balance
that emphasizes solo parts, added reverb, EQ, etc.). In other
words because they were allowed to do the recording in the way
they believed was right, usually employing very simple recording

If this is true, it means we are being deprived of good recorded
sound not because the technology isn't performing, but just
because bad sound sells better.



Laboratorio di Acustica ed Elettroacustica
Parma, Italia

Lascia la spina, cogli la rosa.

More information about the Linux-audio-user mailing list