Re: lame inserting 0.047 seconds of silence in the beginning of the mp3 file

> "so your problem is solved by decoding externally for now and in the long
> term have audacity use libmpg123 (or do the needed work themselves on top
> of libmad, but I'd prefere them switching;-)."

Some more explanation: For various reasons, all MP3 encoding and decoding 
involves some partially-predictable number of "junk samples" appearing at the 
beginning and end of the stream. LAME keeps track of this when encoding and 
includes this delay & padding info in the LAME header, which decoders are 
expected to use to know exactly what to trim (along with the trimming they 
should be doing of the junk the decoder adds to the beginning of the decoded 
stream) in order to output a junk-free stream of the exact length as the 
original. The LAME header is embedded in an extra frame of silence (usually 
1152 samples, about .026s) at the beginning of the file; this frame is 
normally skipped by decoders which recognize it as a "VBR info" frame 
containing cue points and other info.

So as you can see, there are multiple possible points of failure that can 
result in extra samples on one end of the file or the other. mpg123 and LAME's 
own --decode do all the recognition and trimming they're supposed to, but I've 
seen popular audio players and editors which decode more naively, not trimming 
all the junk. It's possible this is what is going on in Audacity on your 
system.

------------------------------------------------------------------------------
Virtualization & Cloud Management Using Capacity Planning
Cloud computing makes use of virtualization - but cloud computing 
also focuses on allowing computing to be delivered as a service.
http://www.accelacomm.com/jaw/sfnl/114/51521223/