"Framerate Changing" And Pitch Adjusting With Audacity

[Zavo]

Hi everybody, this is my first post.
I'd be pleased if someone could resolve those little problems that appened to me:

1) Recently I needed to "change the framerate" of an audio stream from 25.000 to 23.976. For this operation I used BeSweet. I'd like to know how to use Audacity to perform the same operation; I suppose I have to use Audacity's effect "change speed", with a shifting percentage of -4.096. Is my hypothesis correct?

2) After the operation above, the pitch is alterated as a consequence of the speed reduction. I used the "change pitch" effect of audacity to restablish the original pitch (with a the percentage of shifting of 4.096). Here comes the trouble: after the application of the audacity's "pitch change" effect, the quality of the audio stream get worse, or better, different. In the specific, there is a strange reverberation, the stereo effect "get crazy" and "wider" with voices moving strangly from a channel to the other (sorry for the inaccuracy of my description but I'm a newbye in audio editing)...Does it exist a solution to this issue?

[Zavo]

The steps 1) and 2) can be performed with only one operatione, using audacity's effect "change tempo" (low the tempo of a percentage of -4.096). Unfortunately, the audio file produced is still afflicted with the "wide and crazy" stereo effect.
Could anyone solve this please?

[LigH]

I would just guess that the speed change (audio does not have frames) uses filters with a rather low quality. Is BeSweet not suitable for you?

[Zavo]

Originally posted by LigH
I would just guess that the speed change (audio does not have frames) uses filters with a rather low quality. Is BeSweet not suitable for you?

No, because I specify that the lost of quality happens after the step 2 (pitch changing). I have no problem in step 1.
BeSweet in my case is not suitable 'cause it doesn't allow to adjust the pitch (I think)...

[LigH]

Originally posted by Zavo
BeSweet in my case is not suitable 'cause it doesn't allow to adjust the pitch (I think)...

Then please read the OTA options again:

-r [destination] [source]
---------------------------

Default: not used

This sets Frame-Rate-Convertor ratio.

common values :
for NTSC2PAL (23.976fps to 25.000fps) use "-r 23976 25000"
for NTSC2PAL (29.970fps to 25.000fps) use "-r 29970 25000"
for PAL2NTSC (25.000fps to 23.976fps) use "-r 25000 23976"


And look at the middle left part of the BeSweetGUI, at the top of the group "OTA (Overall Track Adjsutments) Options".

(Yes - the typo is intentional: DanniDin, please fix!)

[Matthew]

BeSweet doesn't adjust the pitch, that is well known.

Pitch correction is troublesome, and the only app that people have bragged about is Prosoniq Timefactory. So I would expect Audacity's pitch correction to make the audio stream worse.

If the original source was film, it's likely that the 24->25 fps conversion for the audio was done without pitch correction. Hence in any event it may be better not to undertake any pitch correction. Simply by stretching the audio you are getting the original pitch back.

[SomeJoe]

Sonic Foundry (now Sony Pictures) Sound Forge will do time stretch/compress with built-in pitch compensation, and I've had very good results for compression and stretches less than about 8%.

[Zavo]

@LigH: please forgive me but I can't understand the meaning of your answer. As I clearly explained above, I have already used BeSweet to do a "framerate changing" operation of an audio stream with success, using the options you mentioned. My question wasn't "can BeSweet perform that operation?", 'cause I know It can. I was wondering how to do the same with Audacity, but only for "technical curiosity".

My major concern was about the "pitch adjusting" operation (step 2).

@Matthew: well, in my case changing from 25 to 23.976 cause a significant pitch lowering.
In the specific, I'm trying to do that:

I've got a video with audio in english. I've got the italian audio too. I'd like to do a double language video, but the italian audio doesn't match to the english video, because it's shorter. I choose to stretchen the italian audio stream 'cause it seems the simplest solution to me, although it requires a pitch correction.

I'll try to use the others audio editors you suggested and I'll tell you my results.

I apologize if I'm not clear in my explanations, I'm trying to do the best with my rusty english. :)

[LigH]

Please forgive me - I must have misunderstood that. English is not my native language, and there are days I'm so tired... :o

[stephanV]

Originally posted by LigH
I would just guess that the speed change (audio does not have frames) uses filters with a rather low quality. Is BeSweet not suitable for you?

but audio does have frames! a MPEG layer III frame for example consists of 1152 samples. so the term frame rate is not out of place at all :)

more to the point:
i think Matthew is probably right, when speeding up the film to PAL they probably speed up the audio without pitch-correction too, so slowing it down should (probably) result in the original audio. the difference (4%) is less than half a note (half a note is the smallest difference on a notebar in western music), so it shouldnt be that disturbing IMO. however, that is someting everyone has to decide for his or herself of course, but in this case the cure might worse than the discease (as you already have noticed). may you find a good filter for pitch-correction, please let me know :)

[Matthew]

Originally posted by stephanV
but audio does have frames! a MPEG layer III frame for example consists of 1152 samples. so the term frame rate is not out of place at all :)

The framerate is not being changed though, AC3 framerate is 31.25 frames per second, and this is constant regardless of whether video fps is 25 or 29.97. So yes audio does have frames, but this operation does not involve changing the framerate (which was his point). But it's just semantics really :)

[stephanV]

true, but there is a change in framerate lying underneath. the audio is played at a lower sample rate (which is proportional to audio framerate) and then its upsampled to the original samplerate again. it is comparable with adjusting the video framerate (the speed) and then converting it back to the original framerate, like you can do e.g. in virtualdub for video, i think. (not including his pitch-correction step)

as a side note/question: is the AC3 framerate always the same?
audio framerate = samplerate/# samples per frame; so AC3 can only make use of one samplerate then? (im no expert on this)