From: Alister H. <ali...@gm...> - 2014-09-25 07:44:42
|
Hi again, John Cirillo wrote, On 25/09/14 15:40: > Alister, > > Thanks for the patches. I tried the first one on my present home > install of 0.21-11 (still using the old lame 3.97) and confirmed that > now the mp3 files become byte-swapped. > I installed the latest lame 3.99.5 and confirmed that the mp3 files > are encoded correctly. This is a great relief, now I no longer have > to keep using that old lame version. > I'm not sure exactly why this byte swap code was added, but it might > have been to correct something in the older lame versions. I do > remember that everything worked fine until lame got upgraded to 3.98. > I seem to remember that some changes were made to the syntax of the > command line options starting with lame 3.98, so maybe that had > something to do with it. There is this in the LAME changelog, although x86 processors are little endian, so it seems unlikely to me that this is the change (not that I know anything...): LAME 3.98 beta 8 April 13 2008 * o ... o fix for some endianess problem on big-endian machines > But it's still odd that no one else has complained about the hiss > noise problem in these last five years or so. I imagine it is due to these factors: - Not many people use gwc regularly. - A lot of them probably don't export because they process an entire side of an LP in one go, and then use other software to split it into tracks, because GWC isn't the best tool for that. - Some probably export only to ogg. - Some probably found that this feature was broken, but didn't report it (notice the lack of reports in the last 19 months about the export feature freezing gwc!). - Some probably followed the workaround you posted and installed an older LAME. :) > Does that mean that on some systems gwc and the new lame manage to > produce a correct mp3 anyway? Good question. > I wonder if it's sound card related. I don't have time to experiment > with this too much but I know my home and work machines both use an > Ensoniq ES1371. So if the problem is not present for most users, it > would not be prudent to remove that byte swap code in the main gwc > distribution. More needs to be learned about what's happening. I'm on > an i386 architecture so I don't have native byte swapping, so that's > not it. Hmm... > > I went ahead and made the three changes you recommend in 0.21-19 and > so far so good. It didn't freeze up on me and encoded a usable mp3 file. > There is something new in 0.21-19 that I'm not sure I understand. > After I get the "encode to filename" dialog and click OK, I get > another box that says to enter the track name. I didn't know what to > put so I just put the number 1 and it took it, but the box didn't > completely clear and there is no encoding progress bar. > I don't know what it did with that track name? It didn't end up in > the id3 tag info. Good spotting. If you use the "Simple encode selection as mp3" it ends up in the id3 tag. I hadn't got around to comparing the "old" and "new" encode functions yet to look for any other differences like this - I'll do so now. If you're building and using your own copy of gwc anyway, would you mind using my version on github, instead of just this patch? I believe it is less buggy, but testing is good! BTW I haven't had a chance to attempt the last changes needed to finish porting to plain gtk, but hopefully I will soon. > Thanks again, I'm now using the latest lame with no issues. > > John Thanks, Alister |