I have used 3 different front ends to compress WAV files into MP3s using LAME 3.90.3 and the command line --preset standard.

The 3 front ends are dBPoweramp, Razorlame, and EAC. Each front end produces a slightly different file. Why would this be if all 3 are using the same exe file of LAME (3.90.3)???

For the Stereophonics "Since I Told You Its Over"

Razor lame produces a file size 6,641,394
EAC produces a file size 6,501,541
dBPoweramp produces a file size 6,291,976

Why would this be???

I want to make sure I am using the best method to create these MP3s. I prefer dBPoweramp because it can convert my MAC files directly to MP3 without first converting to WAV - therby preserving the ID3V1.1 info. Give that the file size is smaller, am I leaving some quality on the table?

I'd wager that there are variations in the tags on the files. I know that RazorLame does not add tags

As dreamliner77 said the obvious thing to consider are the tags, but it seems strange to me that the Razorlame file isn't the smallest like it should be having no tags at all.

Turn tagging off in all frontends, or remove the tags once encoded and compare again.

My first thought was tags as well, but do tages produce that much in size? Further, as you said - Razor Lame has NO tags and it produces the largest file size. So the "tag is increasing the file size" theory does not work.

Finally, all of these were encoded from WAV files. WAVs store no tag info.

1st: All encodings are done from the same source (same WAV file)? Because if you ripped a song in say Audiograbber, I'm 100% sure the WAV file would not be an exact match of the WAV coming from the same song ripped in EAC (provided you've configured EAC correctly). I seriously doubt Audiograbber has the same degree of "exactness" as EAC does.

2nd: Are you sure you used the exact same .exe of LAME 3.90.3 (and not different compiles)?

Yes. A WAV file created from an APE file (which was originally ripped with EAC). With dBPoweramp, I converted directly from the APE file (it transcodes to WAV then compresses with one button click). With Razor and EAC compress, right from the WAV.

In no case was a CD involved in this latest compression.



Yes. downloaded 3.90.3 last month and copied the same exe into all my directories where LAME is used.

Really weird stuff... Have you tried using lame.exe from the command line? Maybe that will give to the key to know which two/three frontends are "modifying" the file upon encoding.

Are you sure you selected "Only use Custom Options" in RazorLame?

Are you sure you are using the same lame.exe in all frontends?

Even if there are no tags, maybe there's padding for id3v2?

Why don't you decode to .wav (ideally without dither) and do a comparison (using EAC or better a wave editor)?

I seriously doubt that ID3v2 padding would introduce differences in the 100kB range ... very strange thing ...

I would try each copy of the lame.exe from the commandline with the exact wav as a further step or try to delete null samples from the original file before encoding ... maybe some frontend removes them before coding ...

I'm tempted to ask dev0's question again. And to suggest re-checking the command lines passed by each program.

Another thought: are you overclocking your PC?

Cheers,
David.

Doesn't dBPoweramp use a dll for lame?

Hmmm ... just trying to reproduce your problem ... one wav encoded via EAC with 3.90.3, the other encoded with RazorLAME (3.90.3 as well) ... both show exactly the same filesizes.

Maybe your EAC points to another compile and, as Westgrove mentioned, dbPowerAmp MC is using a *.dll ...