ntsc -> pal

[manicminner]

when converting wav audio tracks from an avi ntsc (23,9fps) to a wav for suite with a pal (25fps) video, the resulting audio have a effect similar to a reberveration... is this normal?

witch is the correct way to convert audio from ntsc to pal and maintain it synched with video? (audio part only) :confused:

[DJ Bobo]

For the ... time, if you want perfect results, use Prosoniq TimeFactory.

[DeXT]

CoolEdit Pro will also do the job. It features an audio stretch filter which preserves pitch.

[DJ Bobo]

CoolEdit won't, the result isn't flawless with it.

[Ácaro]

Originally posted by manicminner
witch is the correct way to convert audio from ntsc to pal and maintain it synched with video? (audio part only) :confused:

Hello,

The best way I found out to convert from NTSC to PAL but only from 23.976 to 25 is using BeSweet (freeware program).

But why only 23.976 to 25 and not 29.970 to 25?

Simple (or not :D), when you make this type of conversion two variables are involved, synchronization with video and pitch correction. Well, most of the programs can easily synch audio with video but pitch correction only a few (or one) can.
So, using BeSweet with "Presets: NTSC->PAL (23.976 to 25.000)" your audio is synched with video and the pitch is the same but you don't notice the diference. The sound is clear!


Hope I helped.

[Xesdeeni]

As far as I can tell, the pitch is not correct when converting using BeSweet. Are there any other free alternatives?

Xesdeeni

[DJ Bobo]

Yes exactly, BeSweet (tested version: 1.4) doesn't correct pitch at all.

@ Acaro
There is no point in making 29,97fps to 25fps conversion, because it's simply not reasonable: the difference in speed is too~ high.
When you wanna convert true 29,97fps content to 25, you've got to skip 10 fields to get 24,975fps, then you can accelerate from 24,975 to 25fps.
This is how you skip 10 fields in avisynth:

separatefields()
selectevery(12,0,1,2,3,4,5,6,7,8,11)
weave()
assumefps(25)

Add a blendfields() or fielddeinterlace(full=true) directly after weave() if you want blended progressive output.

[Xesdeeni]

In some cases, you can avoid converting the audio at all. Check out my (under construction) Standards Conversion Page (http://www.geocities.com/xesdeeni2001/StandardsConversion/index.html) and see if that helps.

Xesdeeni

[Ácaro]

@ DJ Bobo

Thanks for the video info.
But, and what about the sound? Do you have a solution for that?
I tried Prosoniq, it worked but crashed after a couple of MB.

[DJ Bobo]

It crashed? may be you should try another version. I have the 1.2 version. On my PC it stops conversion after 30 to 45 minutes depending on the file. Don't know whether the problem is my computer or the software itself (I'm not able to find any newer version :o )
For me, it's not really a problem, because I'm dealing with anime episodes most of the time, and they have less than 25 minutes, so it's not tha~t frustrating.

[Ácaro]

The latest version of Prosonic TimeFactory is 1.61 (PC), you can take a look at Prosoniq (http://www.prosoniq.net/) site. I haven't test it yet, but if you do, please post the results :D

Regards

[DeXT]

Originally posted by DJ Bobo
CoolEdit won't, the result isn't flawless with it. The result IS flawless for me. I may have a worse audition than you but it sounds ok to me, and it's perfectly synched once you enter the right stretch ratio.

And of course it does not stop at 30-45 minutes of processing (what a buggy piece of sh*t, I'd say, but I won't say it as I'm polite enough ;) ).

[bond]

Originally posted by Xesdeeni
As far as I can tell, the pitch is not correct when converting using BeSweet.Originally posted by DJ Bobo
Yes exactly, BeSweet (tested version: 1.4) doesn't correct pitch at all.Does anybody know if there is a pitch correcting in besweet 1.5b, since there were a lot of changes since 1.4?

[DeXT]

Originally posted by bond
Does anybody know if there is a pitch correcting in besweet 1.5b, since there were a lot of changes since 1.4? Nope. Still there's no pitch correction on latest Besweet betas. In fact there was a bug with FRC in recent versions, which has been fixed in the latest one (1.5b16 as of 4/3/03).

I'm willing someone writes a pitch correction plug-in for BeSweet, which used in conjuction with FRC could give perfect results.

[DJ Bobo]

@ Dext
Don't make me laugh. You don't need super ears to detect noise or cracks or distortions. Just put good headphones on your ears and *COMPARE* with the result of Prosoniq TimeFactory.
If you can't find Prosoniq TimeFactory, you can use EZ Time Stretch from the same company. The demo version allows conversion of 10 sec files.

[EDIT]
Just found out that there is a patch in the package. It is for W2K users and should avoid the "not enough diskspace" message. I have W98, but may be it's the same issue. I installed the patch, may be it can handle now longer files. I'll report tomorrow.

[Xesdeeni]

I don't know about a BeSweet plugin, but I'm trying to write one for AVISynth. So far I've been trying the algorithm at www.dspdimension.com. The quality isn't very good, but I'm still not positive it's not in how I'm feeding it the data. It's also unbelievably slow, so if I can get a satisfactory quality out of it, I'm still going to have to figure out how to speed it up, or it will be useless. I may also try with "Ratty Pitch Shift" code in VirtualDub if the DSPDimension code doesn't pan out. And if neither works very well, I'll have to keep searching. So it still is going to take a while....

Xesdeeni

[DJ Bobo]

@ xesdeeni
You better forget it, the programmers of TimeFactory put 8 years research in it :D

[Xesdeeni]

Over 15 years ago, a friend had a Yamaha SPX-90 that could pitch shift in real time. SoundForge does a credible job today. Heck, even my Chinese-made DVD player can pitch shift, even from a CD. It's just a matter of locating a decent algorithm.

Just for fun, I've posted what I've been working on so far at http://www.geocities.com/xesdeeni2001/AVSPitchShift.zip.

For NTSC film (23.976 fps) to PAL film (25 fps), use an AVISynth script like this:LoadPlugin("AVSPitchShift.dll")
AVISource("NTSCFilm.avi")
ConvertToMono()
ResampleAudio(42294)
AssumeSampleRate(44100)
PitchShift(0.95904)
AssumeFPS(25)It only works with mono files, and it's pretty slow (and may contain bugs, so use at your own risk). I also haven't figured out the right parameters to pass to the working portion, or even added them to the parameter list yet. But you can see it's not hopeless. I'm just not happy with the quality or speed...yet. ;)

Xesdeeni

[Xesdeeni]

It's not a plugin for AVISynth (yet), but give the utility referenced here (http://forum.doom9.org/showthread.php?s=&threadid=46576) a try and see what you think.

Xesdeeni

[Imperial Llama]

Are many PAL dvd's actually pitch corrected when the audio speed is changed? I've compared the sound from PAL dvd of Ocean's Eleven with a NTSC dvd-rip and the PAL version was clearly a higher pitch. Changing the pitch of the PAL soundtrack by ~4% seemed to correct the difference (to my ears at least) which would suggest that the no pitch correction had been used for the PAL dvd.

[Xesdeeni]

That's interesting. According to these two links:

http://www.disctronics.co.uk/technology/video/video_film.htm
http://isb.ri.ccf.org/biomch-l/archives/biomch-l-1995-05/00087.html

it's common to correct the pitch. I'm sure older conversions didn't have the technology to do so, but I'm surprised to hear (pardon the pun) that such a new conversion would not have been adjusted.

Xesdeeni

[Imperial Llama]

It's possible that when these articles refer to "pitch correction" they mean correcting the pitch of the audio so it stays in synch with the video, rather then trying to preserve the pitch of the original 24fps sound track. I'd be interested to hear from anyone who knows how the pitch correction is done for PAL conversion or from anyone who is able to compare a NTSC sound track to a PAL one to see if they can hear any pitch difference.

In the past I have always ripped PAL dvd’s at 25fps but I'd change it back to 24fps if correcting the audio is as straight forward as resampling it.

[Xesdeeni]

It's possible that when these articles refer to "pitch correction" they mean correcting the pitch of the audio so it stays in synch with the video, rather then trying to preserve the pitch of the original 24fps sound track.That's not how I read it. Remember that the original conversions from film to PAL were done completely mechanically. Speeding up a 24 fps film to 25 fps also sped up the sound track recorded on the edge of the film. The video and audio were still in synchronization, but the audio was about 4% higher in pitch. I'm sure this really annoyed everyone for a number of years.

I read these two links as adding the capability of today's electronic processing to correct for that pitch change.I'd be interested to hear from anyone who knows how the pitch correction is done for PAL conversion or from anyone who is able to compare a NTSC sound track to a PAL one to see if they can hear any pitch difference.Now I'm confused about your observation above. When you say:I've compared the sound from PAL dvd of Ocean's Eleven with a NTSC dvd-rip and the PAL version was clearly a higher pitch.Do you mean the speed is faster or that the tones are higher? I'm pretty sure that the latter is what most people are referring to when speak of the pitch (http://www.mmk.ei.tum.de/persons/ter/top/defpitch.html) of audio.

But in any case, I also would also be interested in anyone else who can take a listen to both versions of a movie and let us know.In the past I have always ripped PAL dvd’s at 25fps but I'd change it back to 24fps if correcting the audio is as straight forward as resampling it.BeSweet has a built-in setting for the resampling. The previous version of my film conversions used AVISynth to do the conversion (for example, ResampleAudio(Round(44100 * (25 / 23.976))) & AssumeSampleRate(44100) for PAL->NTSC VCD). But in both cases, while the audio remains in sync, the pitch is altered.

WSOLA modifies the duration, which also keeping the audio in sync, but without changing the pitch. The author of the algorithm claims there are no artifacts. In my few tests, I can't hear any. But I'm hoping people with a wider variety of movies and better sound systems than my dinky little computer speakers can share their results.

Xesdeeni

[Imperial Llama]

All I was suggesting in my previous post was that when people talk about changing the pitch of something they might automatically assume/imply that it involves changing the speed too, like the way a "pitch adjust" control on a record deck works. But maybe I've been playing too many records and it's just me who thinks like that. ;)

Originally posted by Xesdeeni
I've compared the sound from PAL dvd of Ocean's Eleven with a NTSC dvd-rip and the PAL version was clearly a higher pitch.
Do you mean the speed is faster or that the tones are higher?

The speed is faster and the tones are higher. I've uploaded some samples so you can check them yourself:

http://varied.myby.co.uk/shared/doom9/ntsc.mp3
http://varied.myby.co.uk/shared/doom9/pal.mp3

I hope that makes it a bit clearer. :)