Subject: Re: [ecasound] possible bug?
From: Kai Vehmanen (kai.vehmanen_AT_wakkanet.fi)
Date: Sat Apr 17 2004 - 17:31:16 EEST
On Tue, 27 Jan 2004, Alexander Klosch wrote:
> my plan was [for a project/performance] to take some binaries, make them
> to wavs and burn them. after all ripping again, making the wavs to
> binaries again and getting what I put in before.
> well, doing like this:
> #> ecasound -i binary.tgz.raw -o binary.tgz.wav
> #> ecasound -i binary.tgz.wav -o newbinary.tgz.raw
> the sizes of all files are equal, but the new binary isnt the same as
> before ..
> doing the same thing with flac works, reversible.
This is more of a feature than a bug. Ecasound always converts audio data
to its internal 32bit-float format before processing (and back when
outputting). If the input/output format uses fixed-point numbers to
represent audio samples, rounding will occur and the processed data won't
be bitwise equal.
You could intrepret the raw data as 32bit-float data, but this doesn't
help as you need to convert to 16bit-fixed before burning to CD.
I recommend using sox or sndfile-convert (from libsndfile package) if you
want to do bit-accurate conversions from one format to another.
--- http://www.eca.cx Audio software for Linux!
This archive was generated by hypermail 2b28 : Sat Apr 17 2004 - 17:26:40 EEST