[LAU] What is the best MP3 encoder?

Peder Hedlund peder at musikhuset.org
Tue Apr 2 12:35:29 UTC 2013


Quoting Ralf Mardorf <ralf.mardorf at alice-dsl.net>:

> On Tue, 2013-04-02 at 12:59 +0200, Peder Hedlund wrote:
>> You should read Monty's (of Ogg/Vorbis and Opus/CELT fame) 24/192 post
>> : http://people.xiph.org/~xiphmont/demo/neil-young.html
>> When recording you should obviously go as high as feasible but 16/44.1
>> is more than adequate for listening.
>
> I didn't read this post, because I already disagree. There already is an
> audible difference between 44.1KHz and 48KHz. Unlikely that my
> previously golden ears are still golden, since I'm living for > 15 years
> in a loud place.

Feel free to post the results of a double blind test to prove you're right.

> If there are no technical issues, e.g. a sound card that does perform
> better at a sample rate > 48 KHz, then to record with more than 48 KHz
> is useless. Why should a sample rate > 48 KHz improve something?

When recording it's (almost) never wrong to go as high as possible,  
since your audio will go through a number of different effects,  
filters and stuff before eventually be summarized on the master bus,  
so you should try minimizing all the potential loss on the way there.
Once you render the file 16/44.1 is all you need - have a look at  
Monty's really good videos http://video.xiph.org/vid1.shtml and  
http://video.xiph.org/vid2.shtml

> And again, no, 44.1 KHz even isn't adequate, you can hear loss of
> quality.

Talk is cheap, prove it in a double blind test please.

> The most important thing is the way people listen to audio signals. Make
> your ABX tests with non-musicians and play two different versions of the
> same song, that only include some differences, e.g. the Bass sound is
> different or even a loud snare always played at 2 and 4 is a different
> sound.

ABX testing isn't about hearing a particular instrument or if the EQ  
is different in one track - it's being able to tell if track X is the  
same as track A or B.
Do try it yourself on a favourite song; it's as easy as encoding it  
with lame, decoding it back to wav with lame, importing the original  
and the original->mp3-new_wav into foobar2000 with the ABX plugin and  
checking if you can hear which is which.
Here's a guide : http://www.youtube.com/watch?v=jt7GyFW4hOI

> And those "certain things" always could be part of the music and usually
> are part of the music, people usually are simply not trained to notice
> this, they wouldn't notice, if Jimi Hendrix plays a Flying V or a
> Stratocaster, the way Hendrix played guitar I suspect every guitarist
> will hear the difference between a Flying V and Stratocaster.

I saw a test where a bunch of professional musicians and engineers  
listened to a guitar player playing an old $5000 Les Paul and a $500  
copy and were asked to tell which was the expensive one. About half of  
them failed, including the guitar player in the group.
The same was true for a Stradivarius and a cheap beginners violin,  
though IIRC the violin player was correct.

  - Peder



More information about the Linux-audio-user mailing list