naturerecordists
[Top] [All Lists]

Re: Question re MP 3 compression; Kherz; and Digital Recordi

Subject: Re: Question re MP 3 compression; Kherz; and Digital Recordi
From: "John Lundsten" lundsten_john
Date: Wed Apr 4, 2007 1:35 pm ((PDT))
 In  "John Lundsten"
<> wrote:

>......also at the time LAME didn't do variable bit rate encoding
which can
> be VERY useful IMO.
>

Rich Peet  Wrote
I don't understand dither well enough to comment and don't understand
various encodes that well either. My comment relates only to natural
sound recording.

I strongly recommend that for natural sound recording, never ever use
Variable Bit Rate MP3. Because even with loud birds there is quiet
between the notes and the result is terrible. On quiet sounds and
changing volume sounds you place a "pumping" sound to your files
background. This can be avoided by using Constant Bit Rate.

I don't do music so I have yet to find a use for Variable Bit rate.

Rich Peet

Wow that's very interesting, Rich do you have an example of the "pumping"  =
I
could hear?
 Eg same original full resolution file with both approaches and with a
similar total data/file size.

Clearly an encode with a max data rate of X will be a whole lot smaller tha=
n
another with a constant data rate of X, the vari Bit rate approach/idea is=

that as some material (Eg recordings with dynamic range,  VBR is pretty
useless for highly compressed pop Music say) it is a lot easier to data
reduce some parts of the original than others  - so where the going is easy=

relatively few bits are used, or needed, but for sudden loud transients tha=
t
can get severely mangled by CBR Mp3 the data rate can temporally shoot up.=

The same applies to stereo width & stability with CBR it goes to mushy mono=

as an attempt to encode tricky stuff when the data rate is inadequate for
the job.
I do lots of sound FX recording (quite a lot of Classical music too) much o=
f
which are very like nature recordings in that they have sections at low
level with a lack of HF transients that are a challenge to a data
compression schemes - then HF rich  and/or very stereophonic events - I've=

found it does what it supposed to do, so I'm interested in your experience=

as it's not what I've found. I can imagine at low data rates this may give=

the nasty effect you mention but for me MP3 is pretty nasty in general and=

apart from mono speech I've little experience of constant (or average) data=

rates below 128k for stereo.

BTW none of this MP3 stuff above has anything to do with Dither
regards
JL






<Prev in Thread] Current Thread [Next in Thread>
Admin

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU