Speed up an bd audio

[phate89]

Hi. I'm trying to speed up an english bd audio track from 23.976 to25 fps to match my pal dvd audio. I was looking for the -speedup function of eac3to because i read is a valid speedup but when i listened the result seems that it doesn't adjust the pitch. Is it true?
In this case can I pipe the audio from eac3to to something capable of adjusting the pitch like sox and pipe it back to eac3to?
Is it possible? And if it is what is the best option? Let eac3to do the speedup and only adjust the pitch in sox (or some other good pitch correction tool) or let sox do everything?
Thanks in advance

[tebasuna51]

@phate89
- When you change the audio duration with eac3to the original pitch is changed.

- You can pipe the eac3to output to sox, but you can't go back to eac3to, you can pipe the output to a encoder:

eac3to <input> stdout.wav <parameters> | sox <parameters> | <encoder> <parameters> <output>

Like <encoder> you can use Aften, NeroAacEnc, etc.

- Let sox do all the job

- Remember than preserve the original pitch work fine for stereo audio, with multichannel audio you can't preserve the phase between channels with free software.

[Music Fan]

Hybrid (which uses Sox and other tools) can change the speed and let the choice to adjust or not the pitch ;
http://forum.doom9.org/showthread.php?t=153035
Here is the way to open an audio file only ;
http://forum.doom9.org/showthread.php?p=1689479#post1689479

- Remember than preserve the original pitch work fine for stereo audio, with multichannel audio you can't preserve the phase between channels with free software.
Interesting, could you explain why this problem happens with 6 channels and not 2 ?
What happens exactly if one change the pitch anyway on a 6 channels track ?

[phate89]

- Remember than preserve the original pitch work fine for stereo audio, with multichannel audio you can't preserve the phase between channels with free software.

My audio is multichannel (5.1). There is no solution to do this with some free software? What is the result of pitch-correct with sox a multichannel audio?

[DoctorM]

I've always used some flavor of BeSweet with SoundTouch for preserving pitch. Usually BeLight.

[von Suppé]

Hi. I'm trying to speed up an english bd audio track from 23.976 to25 fps to match my pal dvd audio.
Can it be that it would be more appropriate/logical to slowdown your pal dvd to 23.976 instead?

...with multichannel audio you can't preserve the phase between channels with free software.

Not even if you would manually unpack each channel to mono-pcm files, stretch/shorten (with or without pitch change) each one the same way, resample to desired frequency, and feed into a (free) surround encoder?
First jump to mind would be that this works for stereo and multichannel. I don't understand why phase between multi-channels can't be preserved.
Can you give any insight on that?

[r0lZ]

Can it be that it would be more appropriate/logical to slowdown your pal dvd to 23.976 instead?Theoretically, it's prohibited. A PAL DVD has to run at 25fps, and converting it to pseudo-NTSC by changing the frame rate only, without resizing the video, is not allowed. It is prohibited too to mix NTSC and PAL content in the same DVD, so you cannot theoretically keep the original PAL menus or bonus. Plus, the NTSC video frame rate is 29.97 fps, not 23.976 fps, that is the "NTSC Film" frame rate, not supported by the DVD specs.
I know that some peoples do that kind of conversion anyway, with sometimes acceptable results, but you can't have the guarantee that the DVD will play in all DVD players. And I suppose that phate89 wants to keep the original audio tracks. IMO, stretching the new audio track is the correct way to do it.

[von Suppé]

Knowing that you are right with all this, r0lZ,
I was wondering if phate89 could play from a device other than a dvd-player (for the matters you describe), or maybe in his case 2 audio only streams matching were to be important - without the video.

Hence the 2nd part of my post concerning shortening the bd audio track.

[phate89]

Can it be that it would be more appropriate/logical to slowdown your pal dvd to 23.976 instead?


Actually no. The Pal ttrack is my native language so is my main audio. English audio is only an extra. I can speed up the bd video without loss of quality but i can't with audio so it's better to preserve the pal language.
Also english usually has higher bitrate than any other language so it has more margin.
Theoretically, it's prohibited. A PAL DVD has to run at 25fps, and converting it to pseudo-NTSC by changing the frame rate only, without resizing the video, is not allowed. It is prohibited too to mix NTSC and PAL content in the same DVD, so you cannot theoretically keep the original PAL menus or bonus. Plus, the NTSC video frame rate is 29.97 fps, not 23.976 fps, that is the "NTSC Film" frame rate, not supported by the DVD specs.
I know that some peoples do that kind of conversion anyway, with sometimes acceptable results, but you can't have the guarantee that the DVD will play in all DVD players. And I suppose that phate89 wants to keep the original audio tracks. IMO, stretching the new audio track is the correct way to do it.

In theory i'm porting the pal audio to the bd remuxing it in a mkv (untouched) since there's no bluray for my language of that movie.
So i bought the english bd and the pal dvd but stretching the english audio is still the best option as i explained to von suppè to keep the best quality of my preferred track, the pal dvd one.

[tebasuna51]

...
Interesting, could you explain why this problem happens with 6 channels and not 2 ?
What happens exactly if one change the pitch anyway on a 6 channels track ?

Was detected in a old thread. I can't find it now.

When there are dialogs in SL-SR channels are out of phase with FL-FR channels and the problem is audible.

The recommended tool to do this process is TIME FACTORY II (http://www.zynaptiq.com/timefactoryii/) (369 €) than "supports phase-synced processing of mono, stereo and multichannel files"
(now Zynaptiq Acquires Prosoniq Product Line and Technologies).

BTW you can try free soft, most the times the problem is not audible.

[Music Fan]

Ok, thanks ;)

[von Suppé]

Was detected in a old thread...

When there are dialogs in SL-SR channels are out of phase with FL-FR channels and the problem is audible.
This raises more questions, for me anyways. I think it has to do with how surround-channels are encoded and --> how decoders work.
For this thread maybe too collaborate to get in-deep to, my guess. Would love to know more though.

[tebasuna51]

This raises more questions, for me anyways. I think it has to do with how surround-channels are encoded and --> how decoders work.

Is not related with encoders/decoders. The same occurs with uncompressed audio (wav file).

Without preserve pitch is only a samplerate transform, and can be exact for all channels, preserve the pitch, or change only the pitch, need a more complex transform not always exact.

[von Suppé]

Remember than preserve the original pitch work fine for stereo audio...
Then I still don't understand why it's possible for 2 channel audio. I know that there are 2 channel audio-cd's that use phase-shifting and other tricks to (make your amp & speakers) create surround sound.
One example that jumps to mind is Roger Water's "Amused to Death" (or am I wrong and is this about something else?).

[tebasuna51]

I supose the soft was write to preserve the phase between 2 channel (standard stereo audio) and can't be modified easily to preserve the phase with more channels. BTW, I don't know how the soft work.

Convert stereo to surround uses controlled phase-shift beteen front and surround channels and never over vocal frequencies, and isn't always perfect. Is not related with the change pith.

[asarian]

Not even if you would manually unpack each channel to mono-pcm files, stretch/shorten (with or without pitch change) each one the same way, resample to desired frequency, and feed into a (free) surround encoder?
First jump to mind would be that this works for stereo and multichannel. I don't understand why phase between multi-channels can't be preserved.

LOL, that''s probably how *I* would do it too. :) If you were to combine the SL-SR channels into a single stereo track, methinks one could change the pitch, no? Same for mono LF channel. And then just splice all 6 channels back into a 5.1 track.

[Music Fan]

As I said above, I believe Hybrid (which uses Sox) can do this with 6 channels tracks, thus you don't need to separate the channels.

[von Suppé]

As I said above, I believe Hybrid (which uses Sox) can do this with 6 channels tracks, thus you don't need to separate the channels.
Ok, I didn't know that. But if it goes wrong, maybe doing each step manually will give you the opportunity to pinpoint which step could be a problem.
Reading all this, I understand that only stretching/shortening audio keeping the same samples is something like, for example in this case, changing the fps header in a videofile from 24000/1001 to 25 in order to just 'tell' the player to play it faster.
The source-samples aren't changed then, but per second more of them are played. Of course this will result in pitch change, like you would let an analog tape run faster.
Stretching/shortening with preserving pitch means that source-pcm binaries go lost and have to be re-calculated and resampled back again to get a common standard (like 48kHz). I think that this is the step (or 2 steps) where things can go not 100% flawless.
I do wonder what order of steps Hybrid uses.

More complex process indeed, as tebasuna stated. I can imagine.
Then it's about changing tones, cords. What to do with timbre and and all those voice- and instruments characteristics? Wow, not easy to re-calculate. I wonder if it can be done 100% flawless, even by industry standard soft/hardware.

[Music Fan]

I understand that only stretching/shortening audio keeping the same samples is something like, for example in this case, changing the fps header in a videofile from 24000/1001 to 25 in order to just 'tell' the player to play it faster.
The source-samples aren't changed then, but per second more of them are played. Of course this will result in pitch change, like you would let an analog tape run faster.
I'm nearly sure this is actually much more complex than just changing the fps in a video stream because only the header change in this case.
While with sound, even when you don't correct the pitch, each sample has to be recalculated because the duration and thus the size of the audio file and the number of samples change, wich is not the case for a video whose fps is changed in header (the number of video frames and the total size don't change).

[tebasuna51]

Of course, if you don't need other audio from the BD, the best option is change the video fps.

That was made with many DVD PAL than is short than the original DVD NTSC.

[TheSkiller]

What if you decode a 5.1 track to 6 mono wavs and then combine FL and SL into a stereo wav and do the same with FR and SR and then apply pitch corrected speed up? Wouldn't that ensure the phase between front and surround channels is kept since the software doing the speed up sees a standard stereo wav?

[von Suppé]

I'm nearly sure this is actually much more complex than just changing the fps in a video stream...

...each sample has to be recalculated because the duration and thus the size of the audio file and the number of samples change...
Aahh of course you're right.
Damn, I knew this and feel stupid to forget about it. Of course changing audiospeed - never mind the pitch - needs resampling whatsoever, if you must stay within same (compliant) samplefrequency.

Thanks for the "heads-up", Music Fan :D

Still (going non-compliant here) I do wonder if it is possible to keep same samples and only change a speed- or samplefrequency header, in any other uncompressed fileformat.

That was made with many DVD PAL than is short than the original DVD NTSC.
I agree. With many dvd's the increase of pitch and speed is clearly audible. Especially with music, 4% is a lot.
What if you decode a 5.1 track to 6 mono wavs and then combine FL and SL into a stereo wav and do the same with FR and SR and then apply pitch corrected speed up? Wouldn't that ensure the phase between front and surround channels is kept since the software doing the speed up sees a standard stereo wav?
Very interesting question, TheSkiller.
When there are dialogs in SL-SR channels are out of phase with FL-FR channels and the problem is audible.
Okay. If you'd unpack the multichannels into uncompressed mono wavs, and do a pitch-preserved stretch after that and pack changed wavs into multichannel FLAC, would that be a way to go?

[Music Fan]

In my opinion that would produce exaclty the same result than doing it in one step on a 6 channels track because anyway each channel will be treated as a mono channel, whatever it's inside of a 6 channels track, except that 6 channels are linked and thus will be gathered (automatically instead of manually) at the end of the process.

[tebasuna51]

I supose is always better process the multichannels than the mono wavs.

At least you have same phase for FL-FR and, I supose than the soft manage pairs of channels, for SL-SR and C-LFE.