fps conversion 25->23,976

[FulciLives]

Hello :)

I have a PAL DVD Rip (25 fps Progressive) that I converted to a 23.976 fps DivX. I created the DivX with no sound using Gordian Knot and VirtualDub (I used Gordian Knot to get the resize and cropping values and then opened the *.avs file in VirtualDub manually and did a 1-pass 100% quality DivX 5.0.2 Pro encode using b-frames).

Anyways, I then took the AC-3 file and using BeSweet (v1.4RC8) tried to convert to a MP3 file while checking the built in option of PAL2NTSC 25.000 to 23.976 option found in the BeSweet GUI (v0.6 b59).

Well, after the conversion/transcode was completed I then muxed the audio with the DivX file using the standard NanDub operation and amazingly the audio is perfectly synced with the image.
HOWEVER ... there is a strange artifact with the sound ... there is a kind of "tapping" noise ... like "tapping" static I don't know how else to describe it. With soft sound passages it either goes away or is very hard to hear but with loud passages it is clearly easy to hear and just not acceptable. I had previously used BeSweet (the stable version 1.3 or whatver it is) without frame rate conversion and the sound turned out just fine, so I am assuming this audio artifact is due to the frame rate conversion.

Is this a known bug or did I do something wrong?
Please any help would be greatly appreciated :)

- John "FulciLives" Coleman

P.S.
Here is the BeSweet log:

BeSweet v1.4RC8 by DSPguru.
--------------------------
Using azid.dll v1.8 (b825) by Midas (midas@egon.gyaloglo.hu).
Using Shibatch.dll v0.2 by Naoki Shibata & DSPguru (shibatch.sourceforge.net).
Using lame_enc.dll v1.28 (4/6/2002), Engine 3.92 <http://www.mp3dev.org/>.

Logging start : 09/25/02 , 23:52:08.

C:\Copy DVD\BeSweetv14RC8\BeSweet.exe -core( -input d:\VIDEO_TS\sleepless AC3 T01 3_2ch 224Kbps DELAY -80ms.ac3 -output d:\VIDEO_TS\sleepless AC3 T01 3_2ch 224Kbps DELAY -80ms.mp3 -logfile C:\COPY DVD\BESWEETV14RC8\BeSweet.log ) -azid( -n1 -c light -L -3db ) -ota( -r 25000 23976 -d -80 -G max ) -lame( --alt-preset 224 )

[00:00:00:000] +------- BeSweet -----
[00:00:00:000] | Input : d:\VIDEO_TS\sleepless AC3 T01 3_2ch 224Kbps DELAY -80ms.ac3
[00:00:00:000] | Output: d:\VIDEO_TS\sleepless AC3 T01 3_2ch 224Kbps DELAY -80ms.mp3
[00:00:00:000] | Floating-Point Process: Yes
[00:00:00:-80] +-------- AZID -------
[00:00:00:-80] | Input Channels Mode: 3/2, Bitrate: 224kbps
[00:00:00:-80] | Output Stereo mode: Dolby surround compatible
[00:00:00:-80] | Total Gain: 0.000dB, Compression: Light
[00:00:00:-80] | LFE levels: To LR -3.0dB, To LFE 0.0dB
[00:00:00:-80] | Center mix level: BSI
[00:00:00:-80] | Surround mix level: BSI
[00:00:00:-80] | Dialog normalization: Yes
[00:00:00:-80] | Rear channels filtering: No
[00:00:00:-80] | Source Sample-Rate: 48.0KHz
[00:00:00:-80] +-------- FRC --------
[00:00:00:-80] | Source Frame-Rate: 25000
[00:00:00:-80] | Dest. Frame-Rate: 23976
[00:00:00:-80] +-------- LAME -------
[00:00:00:-80] | 'abr 224' preset is used
[00:00:00:-80] +---------------------
[01:52:14:736] Gain of 12.0dB had been asserted to file.
[01:52:14:736] Conversion Completed !
[01:52:14:736] Actual Avg. Bitrate : 245kbps
[01:37:40:000] <-- Transcoding Duration
Logging ends : 09/26/02 , 01:29:48.

[DSPguru]

Originally posted by FulciLives
Is this a known bug or did I do something wrong?Koepi also reported this kind of artificats in the PAL2NTSC mode.

[DJ Bobo]

1) why reencoding?! you just have to set video on direct stream copy, set the frame rate to 23,976fps and save to a new AVI, that's it!
2) Time Stretching has to be done in TimeFactory from Prosoniq, which is the *ONLY* (and I mean only!) software that does that kind of processing in studio quality. All other software, including CoolEdit & WaveLab do this in a quite bad fashion.
25 to 23,976fps is stretching by 104,27094%
Check "Polyphonic - best" for best results. You don't have to check "preserve formants" for such small time stretchings I guess.
3) if you would like to display this movie on an NTSC-TV I understand why you would do that kind of conversion. If not, you're wasting your time!

[FulciLives]

Hello :)

To DJ Bobo:
I didn't re-encode the DivX. I made the DivX using Virtual Dub by feeding it an *.avs file I made with Gordian Knot. But I didn't go all the way in Gordian Knot ... in other words I didn't encode it with Gordian Knot I just used Gordian Knot to help me create the *.avs file so that the resolution and cropping would already be set up. I loaded the *.avs into Virtual Dub and created the DivX there, selecting 23.976 frame rate and did a 1-pass 100% quality DivX 5.0.2 Pro (with b-frames) encodinig (at a width of 640 with proper aspect ratio). The DivX turned out perfect. The frame rate changed from the original 25 fps to 23.976 fps and the file was bigger than 1 CD-R but smaller than 2 CD-R Discs (with enough room to mux audio and still be small enough to fit on 2 700MB CD-R Discs). So I only encoded the film once into DivX.

I then used BeSweet GUI (v0.6 b59) along with BeSweet (v1.4RC8) and azid and whatever else etc. to transcode the AC-3 track to MP3 using the built in PAL2NTSC option that BeSweet provides. When I muxed the audio with the previously made soundless DivX (using Nandub to do the muxing in direct stream mode) I was very happy to find that the audio perfectly matches the video.

However, as I said, the audio has a lot of artifacts that make it useless, sort of like a "tapping" or "clicking" sound that is hard to hear on silent or low volume passages but is very easy to hear in loud passages.

To DSPguru:
Since I'm not the only one to report having an audio artifact like this when using the PAL2NTSC option is it safe to say this is a "bug" and will it be fixed in a future release of BeSweet? Is there a work-around of some sort. I posted my BeSweet log perhaps different settings might work better?

- John "FulciLives" Coleman

P.S.
DJ Bobo ... The Prosonic TimeFactory program sounds interesting ... in fact I read another post you made in another thread concerning this program but you do know that it is a very expensive program!
As for wanting to convert from PAL to NTSC it is exactly as you guessed ... to optimize for display on a NTSC TV.

[DSPguru]

Originally posted by FulciLives
To DSPguru:
Since I'm not the only one to report having an audio artifact like this when using the PAL2NTSC option is it safe to say this is a "bug" and will it be fixed in a future release of BeSweet? Is there a work-around of some sort. I posted my BeSweet log perhaps different settings might work better?i'm afraid i didn't have the time to work on this one. no workaround so-far.
sorry..

[DJ Bobo]

@ FulciLives
Check your PM

[FulciLives]

i'm afraid i didn't have the time to work on this one. no workaround so-far.
I just wanted to say that BeSweet kicks ass.
Yes I am kinda upset that the PAL2NTSC option doesn't really work.
But at the same time I want to thank you for creating BeSweet as
it is a much needed well used software program.

So eventhough the PAL2NTSC option doesn't work I still thank you for
BeSweet and will continue to use it unless I need a FRC which is
sometime I will be doing rarely anyways.

- John "FulciLives" Coleman

[DSPguru]

10x john!

[Snakeisthestuff]

i have done an fps conversion from an 25 fps audio file to an 23,976 audio file with besweet and the presets for fps conversion there.
now my audio stream is sync with my video stream but i have an cracking noise all over the audio file.
here is an link for a 10 second example:

http://mitglied.lycos.de/snakeisthestuff/armi3-test.mp3

is there a filter or something else out there to get rid of that cracking noise?

please help :scared:

[Snakeisthestuff]

is nobody here who can help me :(
i think the problem comes trough the sampling rate conversion but, but i dont know how to solve this problem.

[DJ Bobo]

If you searched this forum, you'll have known that no other software than Prosoniq TimeFactory is suited for time stretching. All other software (including semi-pro software like Cool Edit and WaveLab) make noise, crackling or don't do a pitch correction (like BeSweet).
TimeFactory is the only one that gives you perfect results.

Only problem: I noticed that version 1.2 which I have can't deal with long audio files, say after 30 to 45 minutes the conversion will stop.

BTW, what's the big idea converting 25fps to 23,976fps although you're living in Germany, which has PAL standard :confused:

[Snakeisthestuff]

Thx for the hint, and the situation is these:
i have bought armitage 3 in the perfect collection in germany poor quality 2 GB VOB file for 90 minutes :( . so i bought the US THX remastered version, good Movie Quali 4 GB VOB :) , so and now i have an good quali US armitage encode with 5.1 english AC3 sound and want to add the german sound too, and to do this i need to get the german 25 fps audio sync to the 23,976 video stream, and thats why ;) .

[DJ Bobo]

heeee, that's a first! an anime PAL DVD being progressive? wow! never thought it was so...

[airon911]

You can forget Besweet for this task. It produces poor material for me too, so I stick to my own method. It works for material that has been resampled, i.e. it simply plays 4.1666667 percent faster.

Here's how to do it.

Decode with Azid to a 24 bit stereo file(or whatever no.of channels). It takes way more space but it possibly presents you with a better quality file for processing before encoding.

Now you've got a 48kHz WAV(stereo I presume). Get Soundforge ( f*p://files2.sonicfoundry.com/demos/ )and for cryling out loud a kg if all you use it for(anything else, buy it; it's worth it) is doing what comes now. You change the sample rate. You do not resample yet. This is the crucial step realy.

I change the sample rate to 46080(right click on the sample rate at the bottom of the sample window) and then resample back to 48k. That gives you the 24 frame version of the soundtrack. All you need to do is set your video to 24fps after encoding, and mux the wav'ed mp3 of that resampled file :).

For 23.976 (weirdo ratio):

23.976 * 48000 / 25 = new sample rate to resample back to 48000 from

That would be 46033.92 or rather 46034(!). A bunch of audio apps offer this as a working sample rate(Protools for example), so it's well known.

Set the sample rate, resave the file.

Now resample with SSRC. I haven't tried this sample rate yet, but SSRC should manage.
ssrc --rate 48000 source.wav destination.wav

That's it. Hope this helps.

Tony

[airon911]

Btw, normalizing a 24 bit file is way better than doing it to a 16 bit file. Run an exellent limiter over the material instead of using any 'compression' as offered by Azid. Shaving off a dB by limiter sounds way better that any of that compression crap. Take it from an engineer. You can do this in soundforge btw, but test a small piece of material to see if the limiter introduces any kind of delay(the Waves L1 does for example, the L2 doesn't).

That's also the reason I prefer 24 bit output of Azid and dither&noiseshape to a 16 bit file at the end for encoding.

And unless you've got worldclass phase-coherent timestretch software, don't even think about stretching the audio. Only recently has this started sounding good. The PAL LOTR soundtrack unfortunaly sounds rather bad I hear, as collegues have told me. Haven't verified that yet though.

Tony

[DJ Bobo]

@ airon911
Your method isn't good, since it doesn't make pitch correction. Actors become suddenly a slightly different voice.
TimeFactory does a perfect job on pitch correction, so the voices sound the same.

[Snakeisthestuff]

@airon911
thx for your solution, but i have a modem cbc inet connection (3kb/s) and so a 23 mb download isnt a realistic option but if i understand you right you use soundforge to sample the audio at a new rate and this can be done with ssrc too,and with wavefix i can change the wav header information of the sampling rate, but it can only handle 16 bit wav files, better than nothing ;) ,
so if i understand you right i resample my audio to 46034 and then modify the wav file header so that the file will has an sampling rate of 48000 ,ill try it and report back if it works.
btw i have the PAL LOTR Extended Edition and it sounds very good! and it has German DTS too.

@DJ BoBo
i know what your talking about, i will see how much the voice is affected, and maybe try cooledit 2000 cause i see an pitch option there... and yes i would prefer to use TimeFactory if i had this proggy :)

[dar1us]

I got exactly the same problem, sounds wild, but try swapping the source rate with the destination rate, so instead of 25000>23976 - do 23976>25000 - BeSweet (the new one) is mixing them up, when you see BESWEET window actually pop up as it is working, it has got them the wrong way round. Hey,, it worked for me

enjoy


dar1us

[airon911]

@DjBobo:
Timestretching is only desired if the material in fact has been pitch corrected to run 4.1666 percent faster at the same pitch.

In almost all PAL DVDs this is not the case and only now are some pitch corrections starting to sound good enough to be used on full blown movie mixes.

If you know some of the German dubbing voices, you can compare them to their pitch when they are featured in TV series(loads of dub stuff in Germany) or simply compare to a 24 track source such as NTSC if you can. In most cases they are not pitch corrected. People are just used to it here. I don't like it, so I make 24fps versions if I want a DivX version.

The SSRC method is direct reversal of what most 25fps DVDs use these days, IF the source is film.

@Snakeisthestuff:
Nope.
Decode to 48kHz, that's obvious, I know.
Then set the sample rate with any tool that can do this to 46034 Hz.
Then resample to 48KHz with SSRC.
Btw, SSRC can convert from 24-bit to 16-bit and normalize in one go. You may want to try that.

I didn't know there was a tool out there that can do this. Soundforge just happens to be an every day application on my machine so I used that instead. Thanks for the tip. I like commandline automation a whole lot more myself. Now I only need to write a Perl script that can read out the peak value of a file and then apply my limiter to it. The only trouble so far is, that there's no commandline program that can use DirectX or VST plugins yet.

Damn shame. Unattended film encoding is so cool.

Btw, if you wish I can send you a little Perl script I wrote, to find the ~best bitrate for 24fps 1-CD rips from PAL material. You just need to enter the initial frame size(cropped of course) and the number of frames and it spits out video bitrate. The audio bitrate can be set in the script too.

Tony

[airon911]

@dar1us:
Whop. I'll try that. What does it sound like ? Any more pops in the result ?

[dar1us]

I didn't get ANY pops and cracks in the results, this was when I was running b65 GUI of besweet and 1.47 engine... In 1.5 engine (beta) i think it worked fine, cant remember now


darus

[Snakeisthestuff]

@airon911
i dont understand why i have to sample the rate to 46034 and then sample it back to 48000, i thought this didnt change anything or am i so blinded that i cant see that thruth :confused: , ok i have tried it an other way, cause i didnt read your reply yesterday.


my video stream is 92 min 12 secs long
my pal audio stream is 88 min long

so i calculated with your formula the sample rate:
23.976 * 48000 / 25 = 46034, how you said, and then sampled the audio file with cooledit 2000 to 46034. the next step was to change the samplerate info in the waveheader with wavefix back to 48000. then i checked the new length of my audio file, and it was 84 min. so your formula must be wrong for my case of fps conversion.

so i tried 25 * 48000 / 23,976 = 50063,
now i did the same steps described above and get an audio file length of 92 min 12 sec, bingo!

now i muxed audio with video to test sound quali and noticed a strange
thing, at the beginning of the movie audio is sync to video but at 1/3 of the movie it gets slighty out of sync ca -600ms, and at 2/3 of the movie it gets back sync to the movie, i cutted out 600ms at 1/3 of the movie and pasted it at the 2/3 of the movie and now i have a sync german audio stream for my 23,976 video stream, this was alot of trouble :)

and i cant notice any big change in the actors voices so i didnt need any pitch corrections.
the perl script would be nice, never used perl-scripts before :)
but i usually encode HQ and i have never encoded an good 1cd rip before,so i always make min. 2 cd rips. tried Mortal Kombat several times but it all always looks crappy.

@dar1us
thx for the hint, i will try it!
maybe besweet uses the same formula than airon911, this would explain why the preset is mistaken.

[airon911]

Hehe. You got it slightly wrong but got it right in the end.

I do it this way and it may be the same as what you did.

After decode to 48K, I change the information in the WAV header. This is done FIRST.

Then the resampler(SSRC) resamples this audio to 48K. SSRC thinks it's a 46034 Hz file, because the WAV header says so.

The result should be pretty accurate, but the 600ms hoopala may be the result of the strange samplerate.

Why not just run the video at 24fps and use the clean sample rate of 46080 Hz ? That's what I use for my 25 to 24fps conversions.

Tony

[DJ Bobo]

@ airon911
I don't believe it, PAL speed up DVDs not being pitch corrected. I may rather say, almost all of them are pitch corrected. No pitch correction will be just fooling the people. Don't know about TV airings though.

[grug2k]

I've only ever seen pitch correction on one DVD, and that was Lord Of The Rings. And it sucks really bad, there are dropouts and such.

Apparantly the 4-disc version is even worse, with clicks and pops as well, plus being more noticeable due to the higher resolution of the DTS soundtrack. Lucky I bought the R1.

[MvB]

@DJbobo

Where did you ever hear a pitch corrected PAL DVD?
LOTR is the first one i ever heard of. But it doesn't seem to work anyway.

Even if the correction would work it would result in people that are speaking faster than in the original, making it harder to understand for me if i listen to a foreign language like english. I don't know if you know the movie 'Schlafes Bruder' or in english 'Brother of sleep'. I bought it first on Video and partially wasn't able to understand the beginning even the movie was in german which is my native language. Then i bought the american LD and i understood everything excellent (they didn't dubbed it, just added english subtitles):

[airon911]

Most film releases are not pitch corrected. They haven't been doing this, because current algorythems weren't completly phase coherent. Screwing up LOTR will diminish faith here further I fear. I've gotta ask around if there's a good example of pitch correction.

None of the TV stuff is pitch corrected btw and I'm talking about TV series and features made for TV broadcast only(30 fps in the US).

Here the picture is interpolated and gives american material a bit of a washy look. American standards aren't very popular in Europe anyway. We use 48k for standard broadcast material, the US 44.1. Phillips wanted to use 48K for CD material, Sony (or whoever) managed to talk everyone in to using 44.1. NTSC is crap in picture quality compared to PAL. RGB displays are hardly heard of in the US, in Europe everyone uses them for videogames if they can.

Anyway, most material is resampled, i.e. it plays a frame faster on PAL stuff (~4.166666667 percent). People are used to it. No, you can't tell the difference unless you listen for it and know voices very well.

Some DVDs have extra material shot for NTSC TV where the difference is apparent. Toy Story 2 for example also has an extra feature with a full screen version of the outtakes that run at the original speed with interpolated picture.

Tony

[Slogra]

I think some Disney DVDs are pitch corrected. Like Fantasia. Music is really important in this movie, so i guess that's why they corrected the pitch.

I never heard/watched the dvd myself... so i'm not certain.

[Snakeisthestuff]

@airon911
ok i will try it your way.
to the framerate: i use 23,976 cause i do IVTC with the avisynth plugin IVTC.DLL to the 29,970 video stream and, i dont know how to change the framerate to 24, and if i change the framerate to 24, will
there not be inserted a frame somewhere, which makes the avi look a little jerky, yeah i know its only 0.13 so only one frame every 8 seconds will insert :) .

@dar1us
tried your BeSweet hint, it didnt work in my case (Gui 6b65, BeSweet 1.4 final), the presets are right,
25->23,976 = 92 min 12sec (cracking noise)
23,976->25 = 84 min

[dar1us]

try... going to dspguru.doom9.org and nick the latest beta, check it out... but remember not to whine at the flaws, sorry DSP, but it aint perfect... yet ;)


dar1us

[Snakeisthestuff]

@dar1us
tried it with BeSweet 1.5b8 exactly the same results as in 1.4 final.

airon911 method works and with some little cosmetics i got nearly that what i want, so i wanted to say thx to all :)

[airon911]

I'm writing a Perl script to automate the sample rate change in the WAV header right now. Shouldn't be hard, as a friend of mine gave me the details of what to change.

That way, all one needs is to decode with azid to 24 bit, change the sample rate, resample(with normalize and 16 bit dither provided by SSRC) and then encode.

If anybody knows of such a tool to change that small detail in a WAV file, please let me know :).

Tony

[DSPguru]

Originally posted by Snakeisthestuff
@dar1us
tried it with BeSweet 1.5b8 exactly the same results as in 1.4 final.package have been updated. please try again and report back (with logfiles).

[Snakeisthestuff]

@DSPguru
have tried it again with your new package of BeSweet 1.5b8 and BeSweet Gui 6b66. i used 25->23,976 preset.
results:
filelength:
here's something strange, mplayer2 says file goes 85 min and 22 sec and winamp says file goes 92 min and 11 sec (as id should be), both say sampling rate is 48000.

the cracking noise is gone! but the file length?? whats true, what winamp says or what mplayer2 says???

here is the log-file:

----------------------------------------------------------------------
BeSweet v1.5b8 by DSPguru.
--------------------------
Using azid.dll v1.8 (b825) by Midas (midas@egon.gyaloglo.hu).
Using Shibatch.dll v0.22 by Naoki Shibata & DSPguru (shibatch.sourceforge.net).
Using lame_enc.dll v1.28 (4/6/2002), Engine 3.92 <http://www.mp3dev.org/>.

Logging start : 12/14/02 , 00:36:44.

e:\sonstiges\Ripp-tools\Be Sweet\BeSweet.exe -core( -input d:\dvd-ripped\Armitage 3\pal\vts_03_([0x81]_Audio_Deutsch_AC3(2Ch)_48kHz___)_Delay_-40ms.ac3 -output d:\dvd-ripped\armi-out\deutsch-dspguru.MP3 -logfile E:\SONSTIGES\RIPP-TOOLS\BE SWEET\for_dspguru.log ) -azid( -z1 -b1 -s surround2 ) -ota( -r 25000 23976 ) -lame( -m s --abr 128 -p )

[00:00:00:000] +------- BeSweet -----
[00:00:00:000] | Input : d:\dvd-ripped\Armitage 3\pal\vts_03_([0x81]_Audio_Deutsch_AC3(2Ch)_48kHz___)_Delay_-40ms.ac3
[00:00:00:000] | Output: d:\dvd-ripped\armi-out\deutsch-dspguru.MP3
[00:00:00:000] | Floating-Point Process: Yes
[00:00:00:000] +-------- AZID -------
[00:00:00:000] | Input Channels Mode: 2/0, Bitrate: 192kbps
[00:00:00:000] | Output Stereo mode: Dolby surround compatible
[00:00:00:000] | Total Gain: 0.000dB, Compression: None
[00:00:00:000] | LFE levels: To LR -INF, To LFE 0.0dB
[00:00:00:000] | Center mix level: BSI
[00:00:00:000] | Surround mix level: BSI
[00:00:00:000] | Dialog normalization: No
[00:00:00:000] | Rear channels filtering: No
[00:00:00:000] | Source Sample-Rate: 48.0KHz
[00:00:00:000] +-------- FRC --------
[00:00:00:000] | Source Frame-Rate: 25000
[00:00:00:000] | Dest. Frame-Rate: 23976
[00:00:00:000] +-------- LAME -------
[00:00:00:000] | Bitrate method : ABR
[00:00:00:000] | Avarege Bitrate : 128
[00:00:00:000] | MP3 Min bitrate : 32
[00:00:00:000] | MP3 Max bitrate : 320
[00:00:00:000] | Channels Mode : Stereo
[00:00:00:000] | Error Protection: Yes
[00:00:00:000] +---------------------
[01:28:24:800] Conversion Completed !
[01:28:24:800] Actual Avg. Bitrate : 123kbps
[00:22:45:000] <-- Transcoding Duration
Logging ends : 12/14/02 , 00:59:29.
----------------------------------------------------------------------

[airon911]

Haven't had time to do a full blown test yet, but upon doing a quick one this afternoon, Besweet crashed on me and put out a sonically empty file.

Probably some whacked out parameters on my part, but I'll conduct further tests on some material tonight. Of course, I'm interested in the 25->24 conversion most.

Btw, any more thoughts about a limiter instead of 'boost' compression ?

A question. What's the bit depth AZID decodes to, before normalizing is done ?

Take care
Tony

[DSPguru]

Originally posted by Snakeisthestuff
here's something strange, mplayer2 says file goes 85 min and 22 sec and winamp says file goes 92 min and 11 sec (as id should be), both say sampling rate is 48000.your attention to q22 :) :
http://forum.doom9.org/showthread.php?s=&threadid=7633

the cracking noise is gone! but the file length??o, yea !

Originally posted by airon911
Haven't had time to do a full blown test yet, but upon doing a quick one this afternoon, Besweet crashed on me and put out a sonically empty file.please post your logfile.

Btw, any more thoughts about a limiter instead of 'boost' compression ?currently, none.

What's the bit depth AZID decodes to, before normalizing is done ?32bits fp.

[Snakeisthestuff]

@DSPguru
he he all clear now :) thx.

[airon911]

I'm not shure I've used the parameters right the first time. This first try resulted in an MP3 that was overshooting all the time. I took a look at the decoded WAV file and it was way over the top. Here's the log of that first transcoding.
---------------
BeSweet v1.5b8 by DSPguru.
--------------------------
Using azid.dll v1.8 (b825) by Midas (midas@egon.gyaloglo.hu).
Using Shibatch.dll v0.22 by Naoki Shibata & DSPguru (shibatch.sourceforge.net).
Using lame_enc.dll v1.28 (4/6/2002), Engine 3.92 <http://www.mp3dev.org/>.

Logging start : 12/15/02 , 10:26:40.

v:\video_ts\besweet\BeSweet.exe -core( -input v:\video_ts\40days40nights AC3 T01 3_2ch 384Kbps DELAY -80ms.ac3 -output f:\Temp\40days40nights AC3 T01 3_2ch 384Kbps DELAY -80ms-New.mp3 -logfile V:\video_ts\besweet\BeSweet-40d.log ) -azid( -n1 -L 0dB --maximize ) -ota( -r 25 24 -d -80 -G 1 ) -lame( --alt-preset cbr 160 )

[00:00:00:000] +------- BeSweet -----
[00:00:00:000] | Input : v:\video_ts\40days40nights AC3 T01 3_2ch 384Kbps DELAY -80ms.ac3
[00:00:00:000] | Output: f:\Temp\40days40nights AC3 T01 3_2ch 384Kbps DELAY -80ms-New.mp3
[00:00:00:000] | Floating-Point Process: Yes
[00:00:00:-80] +-------- AZID -------
[00:00:00:-80] | Input Channels Mode: 3/2, Bitrate: 384kbps
[00:00:00:-80] | Output Stereo mode: Dolby surround compatible
[00:00:00:-80] | Total Gain: 16.822dB, Compression: None
[00:00:00:-80] | LFE levels: To LR 0.0dB, To LFE 0.0dB
[00:00:00:-80] | Center mix level: BSI
[00:00:00:-80] | Surround mix level: BSI
[00:00:00:-80] | Dialog normalization: -4dB
[00:00:00:-80] | Rear channels filtering: No
[00:00:00:-80] | Source Sample-Rate: 48.0KHz
[00:00:00:-80] +-------- FRC --------
[00:00:00:-80] | Source Frame-Rate: 25
[00:00:00:-80] | Dest. Frame-Rate: 24
[00:00:00:-80] +-------- LAME -------
[00:00:00:-80] | 'cbr 160' preset is used
[00:00:00:-80] +---------------------
[01:31:26:704] Gain of 16.5dB had been asserted to file.
[01:31:26:704] Conversion Completed !
[01:31:26:704] Actual Avg. Bitrate : 166kbps
[01:12:57:000] <-- Transcoding Duration
Logging ends : 12/15/02 , 11:39:37.
--------------------------

The second try turned out fine. I switched off the dialog normalization and the post gain. The normalizing gain set to 0.98. Here's the log.
--------------------------
BeSweet v1.5b8 by DSPguru.
--------------------------
Using azid.dll v1.8 (b825) by Midas (midas@egon.gyaloglo.hu).
Using Shibatch.dll v0.22 by Naoki Shibata & DSPguru (shibatch.sourceforge.net).
Using lame_enc.dll v1.28 (4/6/2002), Engine 3.92 <http://www.mp3dev.org/>.

Logging start : 12/15/02 , 14:48:31.

u:\Programs\GordianKnot\Besweet\BeSweet.exe -core( -input v:\video_ts\40days40nights AC3 T01 3_2ch 384Kbps DELAY -80ms.ac3 -output f:\Temp\40days40nights.MP3 -logfile U:\Programs\GordianKnot\Besweet\BeSweet-40e.log ) -azid( -g 0,98 -L 0dB ) -ota( -r 25 24 -d -80 ) -lame( --alt-preset cbr 160 ) -profile( AIron Movie Decode )

[00:00:00:000] +------- BeSweet -----
[00:00:00:000] | Input : v:\video_ts\40days40nights AC3 T01 3_2ch 384Kbps DELAY -80ms.ac3
[00:00:00:000] | Output: f:\Temp\40days40nights.MP3
[00:00:00:000] | Floating-Point Process: Yes
[00:00:00:-80] +-------- AZID -------
[00:00:00:-80] | Input Channels Mode: 3/2, Bitrate: 384kbps
[00:00:00:-80] | Output Stereo mode: Dolby surround compatible
[00:00:00:-80] | Total Gain: 12.646dB, Compression: None
[00:00:00:-80] | LFE levels: To LR 0.0dB, To LFE 0.0dB
[00:00:00:-80] | Center mix level: BSI
[00:00:00:-80] | Surround mix level: BSI
[00:00:00:-80] | Dialog normalization: No
[00:00:00:-80] | Rear channels filtering: No
[00:00:00:-80] | Source Sample-Rate: 48.0KHz
[00:00:00:-80] +-------- FRC --------
[00:00:00:-80] | Source Frame-Rate: 25
[00:00:00:-80] | Dest. Frame-Rate: 24
[00:00:00:-80] +-------- LAME -------
[00:00:00:-80] | 'cbr 160' preset is used
[00:00:00:-80] +---------------------
[01:31:26:704] Conversion Completed !
[01:31:26:704] Actual Avg. Bitrate : 166kbps
[01:10:29:000] <-- Transcoding Duration
Logging ends : 12/15/02 , 15:59:00.
-------------------------
Any ideas for optimization of these settings ? As an engineer in post production, I kinda prefer slightly higher bitrates so complex stuff doesn't drop too ahrshly in quality. I use CBR mp3 because VBR doesn't sync up properly and isn't supported well enough(so I'm told). Any ideas towards that ? How does OGG do in this respect ? I tried it once, but it just blew up in my face sync-wise.

In any case, thanks for a great tool once again. Thumbs up all the way this time(on my 2nd try:).

Tony

[Snakeisthestuff]

i think exactly as you, i want hq sound, so i ad directly the AC3 file if i could, if not i use ogg where i could, its a great audio format and i have no sync probs with it since koepi's oggmux, you could also use virtualdub mod, or use Cyrius commandline based tools they are working great too.

[DSPguru]

yes, your first try included wrong combination of gain assertion (you asserted it twice : azid+postgain).

your second commandline (made with BeSweetGUI :) ?) is fine.
you might also want to try the commandline :
BeSweet.exe -core( -input .. -output .. -logfile .. ) -azid( -L 3dB -c normal ) -ota( -hybridgain -r .. .. -d .. ) -lame( .. )

ps,
Vorbis is more advised than MP3.

[airon911]

Thanks.

I think I'll try OGG. At the back of my head I have the notion of playing the files on standalone players some time in the future. Not all that realistic I must confess, since I go for 24 fps playback these days, and there's no sync'ing TV display for that frequency :).

-L 3dB would actualy boost the LFE channel by 3dB. Any specific reason or do you just enjoy feeding a sub woofer ? Imrovements on the encoded file later on ?

The hybridgain kinda scared me away, as I wasn't shure what it would do when.

I like milestones like this, where's it's only up to the user now :). Everything works and is in place.

'cept what people like come up with.
When I use Lame to produce a mono mp3, I get an empty mp3 or the correct length but with no sonic information. I have to write a stere WAV and then encode that with the '-m m' parameters in Lame.

If some else can verify it ?

@snakeisthestuff
I'll give OGG another spin.