The No.1 Website for Pro Audio
Help needed with 24fps to 25fps Pitchcorrected 5.1
Old 7th February 2020
  #1
Lives for gear
 
DSK's Avatar
Help needed with 24fps to 25fps Pitchcorrected 5.1

Hey there post wizards!

I'm delivering the sound of a trailer that I mixed in 5.1

The source video was 24fps and I need to make also a broadcast/digital distribution version at 25fps


Usually I just supply the film version so this is a bit new to me

They specifically asked for 25 fps pitch corrected versions.
Does this mean that I have to shrink it to size while preserving the original pitch?


I plan to use Ircam TS on highest settings as it seems to preserve all the transient information properly but I just need to make sure that I'm actually going about this right.

I know conversion questions pop up from time to time, just asking as I got even more confused when I read all the other threads.



Thank you for listening!
Old 8th February 2020
  #2
Lives for gear
Yes, I believe, they want shorter version of audio track to play in sync with 25 fps video clip.
Generally that means, they take a video track and modify it to play frame by frame at ~ 4% higher speed.
Of course audio track is out of sync then. So you need to speed it up accordingly. If you do that in a simplest way (eg. varispeed with sample rate conversion), you'll have no artifacts, but everything musical will be sharp and out of tune, hence they asked for the pitch-corrected version.
So you need to time stretch that audio clip with some rather elaborate algorithm to fulfill that demand.
(so stretch factor is 24/25 = 0.96 and conversely playback speed factor as required by some tools is 25/24 = 1.0416 = 104.16 %)

It looks simple, but with complete mix it can be tough process, because the algorithm needs to handle both harmonic and transient parts of signal with least amount of perceivable artifacts (modulation, flanging like effects, grainy and stuttering sounding sustained tones, veiled natural sound textures, altered imaging in multi-ch). The other complication is with surround, as just handful of tools can do that natively.
You can certainly try Ircam TS for that, but as I've tested it couple years ago (after initial release 1.0), it didn't seem to be right tool for such kind of jobs. Its forte is apparently in real-time manipulation with extreme amounts of pitch correction or brutal time stretches for experimental and sound design stuff with separation of signal components, but it far from invisible (ideal there) with complex mixes.

Michal
Old 8th February 2020
  #3
Lives for gear
 

Hi,
Before you start (if you haven't already), make sure of the process used to do the 24>25fps picture conversion.

It's quite possible that they used the Teranex process to maintain exact speed of action, in which running time remains the same and therefore no speed/pitch change would be needed.

Good luck
Old 8th February 2020
  #4
Lives for gear
 
gsilbers's Avatar
 

Quote:
Originally Posted by DSK View Post
Hey there post wizards!

I'm delivering the sound of a trailer that I mixed in 5.1

The source video was 24fps and I need to make also a broadcast/digital distribution version at 25fps


Usually I just supply the film version so this is a bit new to me

They specifically asked for 25 fps pitch corrected versions.
Does this mean that I have to shrink it to size while preserving the original pitch?


I plan to use Ircam TS on highest settings as it seems to preserve all the transient information properly but I just need to make sure that I'm actually going about this right.

I know conversion questions pop up from time to time, just asking as I got even more confused when I read all the other threads.



Thank you for listening!

the above is good advice.


a few things. make sure you are really in true 24fps and not 23.98.

an easy way to understand it is that these numbers (23.98, 24, 25) are actually names. names that mean what speed the movie will be played at.
23.98 and 29.97 mean NTSC or video speed. 25fps means PAL speed. 24/30 is film speed. there are 3 main speeds. its a simplified audio version of course.

the different is about 4% faster in pal so pal files will be shorter. and since it goes faster the pitch will change so a pitch change is needed. its about semi tone.


i dont understand why distributors ask for this pal audio file without a video reference but if you can , ask for a pal video reference. that way you can double check its in sync. its very ok to ask for a reference in pal.


in pro tools there are presets. those are the one most post prodcution houses use. you import/varispeed into a new session selecting 24 to 25. or 23.98 to 24 and then to 25. or if its true 24 then just 24film to 25pal speed conversion.
quality is more than good. ive done probably close to 5,000 of these and protools is fine so dont sweat it. check it of course. and compare with your other app/plugin.
Old 8th February 2020
  #5
Gear Nut
 

Quote:
Originally Posted by DSK View Post
25 fps pitch corrected versions.
It's not Teranex then. You'll have to speed up & repitch.
The Protools SRC import is a varispeed process: dead accurate on sync, but at the wrong pitch.
Especially on long form, I always had best results separating the two processes: first pitch change using PT Audiosuite (allowing to check sync of pitched audio), then varispeed using SRC import.
Be sure to check/correct start beep, if any.
Old 8th February 2020
  #6
Lives for gear
BTW @ DSK What DAW and platform you're using?

Michal
Old 8th February 2020
  #7
Lives for gear
 
DSK's Avatar
First of all thanks for the answers guys you've helped me a ton once again!

Quote:
Originally Posted by msmucr View Post
Yes, I believe, they want shorter version of audio track to play in sync with 25 fps video clip.
Generally that means, they take a video track and modify it to play frame by frame at ~ 4% higher speed.
I think I finally realised why a second of audio in 24fps is not a second of audio in 25fps... It's because of the process...

I.E. if you have 24 frames x 4 seconds = 96frames/ 4 seconds
But to have 25 frames x 4 seconds = 100 frames/ 4 seconds from the same video you need to take get an extra 4 frames from the next second of the original 24fps...

so literally it's just a faster playback rate? Damn I feel silly now


Quote:
Originally Posted by msmucr View Post
It looks simple, but with complete mix it can be tough process, because the algorithm needs to handle both harmonic and transient parts of signal with least amount of perceivable artifacts (modulation, flanging like effects, grainy and stuttering sounding sustained tones, veiled natural sound textures, altered imaging in multi-ch). The other complication is with surround, as just handful of tools can do that natively.
You can certainly try Ircam TS for that, but as I've tested it couple years ago (after initial release 1.0), it didn't seem to be right tool for such kind of jobs.
I want to try with TS because it natively converts multichannel tracks in one go. I'm afraid that my 5.1 upmix will be ruined by processing it as separate mono, phasing, correlation and all that jazz you've mentioned.

After not liking the algorithm at first I played around with the settings some more and it felt amazingly accurate and transparent on a previeous 5.1 MSX track I've done. I'll try it on the full mix as well and see how it goes!
Old 8th February 2020
  #8
Lives for gear
 
DSK's Avatar
Quote:
Originally Posted by msmucr View Post
BTW @ DSK What DAW and platform you're using?

Michal
I'm making the original recording and production in Ableton Live as it's amazing for sound design.

It's also great for composing, recording and editing 40 seconds high end cues from scratch in half a day (which unfortunately happens more often than I'd wish as the cut progresses)... such is the condensed nature of trailer creation.

For long form scoring I'd rather work in Cubase as I love it's handling of sample libraries and the customization options there regarding midi etc..


After that I'm exporting stems into Reaper for mixing/upmixing 5.1 + delivery template.

It just saves me a ton of time and spits all necessary combinations . Usually after the green light on the final cut I have only a couple of days to provide the 5.1 mix and full deliverables, and between our internal QC this means there's no time to do a fully fledged 5.1 mix unfortunately.


I tried the same in Cubase/Pro Tools but I'm not fast enough, or skilled enough to properly pull it off in such a short time.

Our material has been broadcasted all over EU and so far our NA releases have passed the QC at very reputable labs so I'm not thinking of changing this flow for now.

I actually stopped my ProTools subscription as I was only using it 2-3 times a year and it made no financial sense to keep it going.
Old 8th February 2020
  #9
Lives for gear
 
DSK's Avatar
Quote:
Originally Posted by gsilbers View Post
a few things. make sure you are really in true 24fps and not 23.98.
.......
in pro tools there are presets. those are the one most post production houses use. you import/varispeed into a new session selecting 24 to 25. or 23.98 to 24 and then to 25. or if its true 24 then just 24film to 25pal speed conversion.
quality is more than good. ive done probably close to 5,000 of these and protools is fine so dont sweat it. check it of course. and compare with your other app/plugin.
I'll ask our editor what the source is!
Thanks for simplifying NTSC PAL and film in relation to audio appreciated

Don't use ProTools that much these days, will probably resub just to check the quality of the conversion. Thanks for the pointers!

Quote:
Originally Posted by gsilbers View Post
i dont understand why distributors ask for this pal audio file without a video reference but if you can , ask for a pal video reference. that way you can double check its in sync. its very ok to ask for a reference in pal.
Quote:
Originally Posted by JoeMilner View Post
Hi,
Before you start (if you haven't already), make sure of the process used to do the 24>25fps picture conversion.
Quote:
Originally Posted by FlorianE View Post
It's not Teranex then. You'll have to speed up & repitch. The Protools SRC import is a varispeed process: dead accurate on sync, but at the wrong pitch.
Especially on long form, I always had best results separating the two processes: first pitch change using PT Audiosuite (allowing to check sync of pitched audio), then varispeed using SRC import.
Be sure to check/correct start beep, if any.
We provide the deliverables for the DCP Lab so I'll have to check with our editor and see what kind of PAL conversion he does.. but I know we don't have a Teranex... might look into getting one of those boxes if it can help keep my audio intact across formats
Old 8th February 2020
  #10
Lives for gear
Hi and you're welcome!

Quote:
Originally Posted by DSK View Post

I think I finally realised why a second of audio in 24fps is not a second of audio in 25fps... It's because of the process...
...
so literally it's just a faster playback rate?
Yes, it's faster playback rate.
In general when you want to deliver a clip or movie in other rate than original, you need to alter either video or audio, there's no other chance.
If you alter video (eg. passing it through some transcoder software or mentioned Teranex hw boxes), then it can affect video quality.. depending on used method, you might end up with blended original frames or artifacts in complex scenes (when methods with motion estimation are used), in any case it isn't lossless process and you really need to QC every scene, especially if it goes to some bigger screens (of course you can find lot of projects, where it is not much issue and are routinely transcoded in real-time.. some TV stuff etc.).
However if you decide to avoid that, you take the video frame by frame, adjust playback speed without any modifications to frames, however then you just shifted your issues to audio track, which needs to be time stretched.

Of course in some projects you know right from the start, the procedure happens. In general, the later you do such stretching, the less control over process you have. So if you know that in advance and have time/budget for that, you can avoid doing that with full mix and for instance do two different mixdowns from individually stretched stems. So you can fine tune stretch algorithms for dialog tracks, SFX, different music parts etc. very often with better results.

Quote:
I want to try with TS because it natively converts multichannel tracks in one go. I'm afraid that my 5.1 upmix will be ruined by processing it as separate mono, phasing, correlation and all that jazz you've mentioned.

After not liking the algorithm at first I played around with the settings some more and it felt amazingly accurate and transparent on a previeous 5.1 MSX track I've done. I'll try it on the full mix as well and see how it goes!
Definitely try it and be sure to compare it to some other contenders. As I've said, it was initial release I tested, but it didn't convince me for complete mix processing as opposed to individual elements and sound design.. YMMV.


Quote:
Originally Posted by DSK View Post
I'm making the original recording and production in Ableton Live as it's amazing for sound design.

It's also great for composing, recording and editing 40 seconds high end cues from scratch in half a day (which unfortunately happens more often than I'd wish as the cut progresses)... such is the condensed nature of trailer creation.

For long form scoring I'd rather work in Cubase as I love it's handling of sample libraries and the customization options there regarding midi etc..


After that I'm exporting stems into Reaper for mixing/upmixing 5.1 + delivery template.
Well if you're going to Reaper for upmixing, then you can do time stretching there and rather very well (IMO better than with Ircam TS).
It uses latest Elastique Pro algo, which one of best and handles natively also multichannel audio.
Also there is another workflow question. If that's possible, you can time stretch 2ch audio before upmixing.
Of course if there is also automation (as opposed to some static "print" template) you'll need to solve that.
Reaper automation is possible to stretch as well if you need, there are few ways for that. One of them is conversion of normal automation to automation items, which are then easily stretchable.. (you can simply drag edge of automation item and it snaps to edge of already shortened audio item).

Anyway, if you plan to try Elastique in Reaper, I'd recommend to import complete multichannel clip (eg. poly wav) or stereo wav. Open item settings (F2) and enter desired playback rate (eg. 1.041667), then select algo for Elastique Pro 3 (or 2.2 for test) and under its options, select synchronized mode. That's really important for phase coherence.
After you apply that, you'll trim right edge of the item (as you've just increased playback speed, but item length stayed the same, there is leftover, so you need to shorten it).
Then you can use Batch/file item converter, add selected item there and export it to new wave.

Another important thing is, the stretch process (regardless of tool) really deconstructs original audio and creates completely new samples.. Thus there can be (even up to 6dB for certain situations) increase of max. peak level. So keep that in mind and either give it some headroom before process by attenuation or be sure, any possible intermediate files are in floating point format (with encoded values over digital zero). After you do the stretch you can always use some limiter to bring it back to original levels or just limit newly created peaks.

My final tips there for comparison are Izotope Radius (say in RX7) and ZTX by Zynaptiq.
ZTX algorithm was formerly called Dirac. I always liked that as one of top algos for mix pitch/time stretching.. often above anything what I tried (especially with mixed content.. dialogs, modern percussive or electronic music and some symphonic passages.. that's always difficult).
Their standalone fully featured Time Factory app is cool, but bit pricey.
However the algo is included in some other DAWs. It is at recent Wavelab or MOTU DP, but neither of them handle multichannel wavs AFAIK.
But it is also included in TwistedWave editor . So if you have access to some Mac, definitely try that.
It handles multichannel files well (with phase coherent stretching) and there is fully featured 30 day demo, including saving .
IMHO that algo alone is worth of admission, even if you won't use it for anything else.

Michal

Last edited by msmucr; 8th February 2020 at 11:55 PM.. Reason: left -> right edge
Old 8th February 2020
  #11
Lives for gear
 

everything has already been said before, just one thing:

Don't do the speed up without a reference video (which HAS to be the/a direct derivate from the final 25fps video) with the new frame rate provided to you!
Old 8th February 2020
  #12
Lives for gear
 

Quote:
Originally Posted by FlorianE View Post
It's not Teranex then. You'll have to speed up & repitch.
You're probably right, but it's also possible that the distributor simply provided the OP with their boilerplate deliverables list with no idea how the picture conversion was done- all they know is that they want it in sync and at original pitch

Judging by all the replies, yeah, likely this one is NOT Teranexed but I just wanted to raise that flag for others, as prob 2/3 of the 24>25 conversions that I've done HAVE been Teranexed and they're sure a lot easier to do, LOL
Old 8th February 2020
  #13
Lives for gear
 
DSK's Avatar
Quote:
Originally Posted by msmucr View Post
Yes, it's faster playback rate.
In general when you want to deliver a clip or movie in other rate than original, you need to alter either video or audio, there's no other chance.
....
However if you decide to avoid that, you take the video frame by frame, adjust playback speed without any modifications to frames, however then you just shifted your issues to audio track, which needs to be time stretched.
Thanks for the clarifications, I see now why the industry chooses to affect the audio, this particular project really has some striking imagery. Hopefully with more processing and machine learning we can get smoother frame generating algos in the future!

Quote:
Originally Posted by msmucr View Post
In general, the later you do such stretching, the less control over process you have. So if you know that in advance and have time/budget for that, you can avoid doing that with full mix and for instance do two different mixdowns from individually stretched stems. So you can fine tune stretch algorithms for dialog tracks, SFX, different music parts etc. very often with better results.
That's a very valid point! I am lucky to have access to all of the individual assets.. I did think of rendering every music part at faster playback rate, but due to time constraints it might not be doable. But I'll break it down as much as I can.

Quote:
Originally Posted by msmucr View Post
Well if you're going to Reaper for upmixing, then you can do time stretching there and rather very well (IMO better than with Ircam TS).
It uses latest Elastique Pro algo, which one of best and handles natively also multichannel audio.

Also there is another workflow question. If that's possible, you can time stretch 2ch audio before upmixing.
Makes total sense, I was having a top down perspective on things
I was thinking that once I completed the upmix I can then stretch...
... but you're 100% right with a bit more time I can generate another upmix from the individually treated stereo stems.

I care a lot about audio quality so I'll jump through any hoop the time allows to get a better result.


Quote:
Originally Posted by msmucr View Post
Anyway, if you plan to try Elastique in Reaper, I'd recommend to import complete multichannel clip (eg. poly wav) or stereo wav. Open item settings (F2) and enter desired playback rate (eg. 1.041667), then select algo for Elastique Pro 3 (or 2.2 for test) and under its options, select synchronized mode. That's really important for phase coherence.
After you apply that, you'll trim left edge of the item (as you've just increased playback speed, but item length stayed the same, there is leftover, so you need to shorten it).
Then you can use Batch/file item converter, add selected item there and export it to new wave.
Awesome tips thank you! I'll get to testing asap!

Quote:
Originally Posted by msmucr View Post
Another important thing is, the stretch process (regardless of tool) really deconstructs original audio and creates completely new samples.. Thus there can be (even up to 6dB for certain situations) increase of max. peak level. ....
..After you do the stretch you can always use some limiter to bring it back to original levels or just limit newly created peaks.

My final tips there for comparison are Izotope Radius (say in RX7) and ZTX by Zynaptiq. .........
Again can't thank you enough for these breakdowns Michal!

I'll get to work!
Old 8th February 2020
  #14
Lives for gear
Quote:
Originally Posted by msmucr View Post
After you apply that, you'll trim left edge of the item.
I'm sorry, my mistake, I've meant right edge to trim item length.

Quote:
Originally Posted by DSK View Post
Again can't thank you enough for these breakdowns Michal!
I'll get to work!
You're very welcome and good luck with your project!

All the best,

Michal
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump