Channels flipped with TooLame..

[MattO]

Hello all,

Encoded a VCD yesterday, the audio was converted as follows:

besweet.exe -core( -input "d:\movie.ac3" -output "d:\movie.wav" -2ch ) -azid( -g max -s surround2 ) -ota( -d -49 ) -shibatch( --rate 44100 )

then I used TMPGenc to encode the AV as a non standard VCD, I configured TMPGenc to use Toolame32 to do the audio, via a .BAT file containing the following line:

toolame32 -s 44.1 -m s -b 128 -p 2 -e -r %1 %2

I then played the disk on my home DVD player through my Marantz amp (DD, DTS & DPL2), and the sound from the front left and front right speakers are the wrong way round :(
A car, on screen, moved from Left to Right, but the sound went Right to Left!
Other than this the quailty is fine.

I checked my system with other disks and it is setup correctly, so something must be going wrong somewhere in the encoding.

can anyone help?

MattO

[DJ Bobo]

May be it's also so on the DVD? if yes, you havn't done anything wrong. Here the solution:
Open the WAV in an audio editing program and switch the right & left channels and save it. Then encode to MP2.
BTW, 128kbps is a way too low, better go for at least 192kbps Stereo (no joint stereo!).

It is not so on the DVD? don't know where the fault is, but here is then my method:
Go to the DOS command window and use azid the following way:
azid -c normal -a movie.ac3 movie.wav
Then open the WAV in TMPG and encode it to MP2.
BTW, There is no real differences in the quality of encoding between TMPG and the other encoders. So you can encode in TMPG without pointing to TooLame, the quality will be roughly the same.

[DSPguru]

Originally posted by DJ Bobo
my method:
Go to the DOS command window and use azid the following way:
azid -c normal -a movie.ac3 movie.wav
Then open the WAV in TMPG and encode it to MP2.you missed the sample-rate conversion step.
not to mention the DPL2 Downmix..
There is no real differences in the quality of encoding between TMPG and the other encoders.come on.. !

[pacohaas]

I believe this is a known issue (http://mikecheng.d2.net.au/layer2/HISTORY) with tooLAME (scroll to the bottom of that page). There has been a switch implemented to switch the input channels for exactly this reason

but the real question is...why do you not use BeSweet to go directly to mp2?

[DSPguru]

i had to change the subject of your thread. sorry, but imho, when L=R & R=L it means that you ended up with a mono signal ;)

[E-Male]

Originally posted by DSPguru
when L=R & R=L it means that you ended up with a mono signal

LOL, the programmers way of thinking :D

[pacohaas]

Originally posted by E-Male


LOL, the programmers way of thinking :D ...is there any other way ;)

[E-Male]

Doc Brown would say that we just don´t think 4dimensional *g*

[DJ Bobo]

@ DSPguru
The sample rate conversion can be done within TMPG ;)
You may be right about that DPL2 downmix, but a DPL downmix is also good enough (of course not as good as a DPL2 downmix through a DPL2 decoder, but still... :D)
About quality: I compared TMPG and i-Media (which is the best MP2 encoder AFAIK) and there was only very slight differences. One was sure, both were bad, I wasn't able to get exactly the same quality as the source even with 384kbps :o (MP2 really sucks!). To think that I thought that MP2 delivers CD quality with 256kbps, man, damn was I wrong! :dodgy:

[DSPguru]

ok, so to your opinion :
ssrc is alike wavelab is alike tmpg.
i-media is alike tmpg is alike toolame.
azid+razorlame is alike floating-point azid.dll+lame+enc.dll
dpl2 is alike dpl.

fine by me :)
but you should be more responsible when you give tips to other forum members.

[MattO]

thank you for the replies,

I was using Besweet 1.4bRC4, but I have never had this kind of problem before with this version.

The reason I use Besweet to encode to WAV and not direct to MP2 is if you use a MP2 as TMPGenc's audio input TMPGenc will still create a temporary WAV file before continuing with its encoding, so giving TMPGenc a WAV file saves Besweet a bit of encoding time.
If anyone knows how to input a MP2 file into TMPGenc without it creating a temporary WAV file then please let me know . . . and I don't mean let TMPGenc encode the 'Video only' then Multiplex them together ;) I was hoping to cut down on the individual processes!

To create my XVCDs I usually use Besweet to encode the AC3 direct to a Surround2 MP2 then encode the video with CCE followed by multiplexing with BBMpeg or Mplex, and the results were very good. But I wanted to try TMPGenc for the video, as, from reading various posts, TMPGenc is meant to have higher video quality for MPEG1 than CCE, hence the reason for the above audio method.
I have used my Toolame32 command line numerous times before, in combination with seperate Azid and SSRC lines, with no problems at all. But I stopped using this technique once Besweet came on the scene with Surround2 :)

128KB and not 192KB : saving Bitrate is obviously very important and, IMHO, 128KB is still very high quality audio, in fact, to my old ears, I can not tell a significant (any) difference between 128KB and higher, obviously I doubt I would go lower than 128KB.

[DJ Bobo]

@ DSPguru
SSRC is like WaveLab but not like TMPG. TMPG doesn't deliver the very best quality downsampling, but still very good, especially down to 44KHz, which is unnoticable even in TMPG.

i-Media may be the best MP2 encoder, but as said, the differences are very marginal, so I won't bother too much. I personally always use i-Media since it's very fast, but I don't wanna throw people here in panic if you understand what I mean ;)

And yeah, Azid+RazorLame is the same as the floating-point xyz solution you compared with :D ... ok roughly the same - I don't wanna make you angry :D
The most important part may be amplifying which is done in floating point in AZID anyway.
If one wanna go even farther, one can add an -F wav24 to the azid command to get higher quality. Here I havn't written that command, since TMPG doesn't accept 24-bit WAV files (but RazorLame aka Lame does). Still one can dither before to 16-bit in WaveLab for example, I know I know... (just wanted to spare a step :D)
I'm aware that there is differences, but those differences are only measurable, not hearable ;)
To be honest, I met just one case, where -F wav24 was obligatory, it was a bad encoded dolby digital track which sounds quite noisy when outputted directly to 16-bit (without dithering)

DPL2 downmix is of course better than DPL downmix, but a DPL2 decoder will do an acceptable work even with a normal DPL downmix, so I don't really care (I don't have a DPL2 decoder btw). AFAIK, a DPL2 decoder can create surround sound even from a stereo source so... IMHO if one really wants surround sound, then just keep 5.1 AC3

@ Matto
You must be very (very) old if you can't detect a difference between 128kbps and 192kbps (I'm talking about MP2 not MP3!). In fact, almost all high frequencies are gone with 128kbps: it sounds damn muffled like a (very) old tape, in fact, the low-pass filter is set very early (somewhere by 12KHz?). I personally never go under 192kbps no matter what, and the higher the better! (as said above, not even 384kbps sounds exactly like the source!)
BTW, 64kbps less for video doesn't mean anything. 64kbps more for audio means in that case (128 -> 192) a huge increase in quality!

[NoX1911]

When i export audio files in Adobe Premiere the wave channels are correct. But when i encode them afterwards with toolame02k (cmdline, "-b 224") the channels will be swapped ("-g" can fix it).
I tested another wave file created with adobe audition and the channels remained the same so i think this affects adobe premiere files only...

winxp
amd
premiere pro (v7)