8mm Forum: Frame by Frame Telecine Sound Syncing?

This is topic Frame by Frame Telecine Sound Syncing? in forum 8mm Forum at 8mm Forum.

To visit this topic, use this URL:
https://8mmforum.film-tech.com/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=1;t=011565

Posted by Janice Glesser (Member # 2758) on June 20, 2017, 02:03 PM:

I just captured a Black Hawk short called Fence Buster - The Story of Babe Ruth.

https://youtu.be/G0rCXV2RC48

I have done sound synchronizing in the past...but the recorded sound seldom matches the video. I use Adobe Premiere for synchronizing and editing... but "Fence Buster" was a bear to do and took hours. Since I can't really change the speed of the voice-over (at least not too much)...I end up cutting up the video and adjusting the timeing on each cut-up segment. Syncing to something like the Babe swinging with the sound of the crack of the bat. The audio must drift in someway over time causing the video to go out of sync.

Is this audio drifting problem attributed to the sound projector or the recording device. I've had better luck with realtime sound and video captures...but not with frame by frame since these files get converted and compressed. Are there any audio capture parameters I should be using? Should I keep trying different sound projectors for better recording?

Any thoughts welcomed [Smile]

Posted by Ty Reynolds (Member # 5117) on June 20, 2017, 04:05 PM:

I now use a scanner that also captures synchronized audio, but this is how I used to do it:

Make sure the film has plenty of leader at the beginning, and at least two inches of leader at the end. Scan and process the film. I was using Moviestuff's Workprinter and Retro8.

Then I run the film again on an Elmo ST1200 with a camera aimed at the projected image. Just a basic point and shoot set-up, the image quality doesn't matter. The video portion will just be a guide track. The audio is fed from the projector to the camera's line in, and both audio and video are fed from the camera to the computer via firewire. This is recorded in my editing program, Vegas. You could also record on tape or card, and then import.

Place the scanned footage on the timeline. Trim the leader to the first frame of film. Trim the end leader to the last frame of film. Calculate the exact number of frames.

Place the second film capture under the first in the time line and trim to the first and last frames. Line up the first frame of each capture. Calculate the exact number of frames in the trimmed camera capture.

The number of frames between the two captures will never be an exact match. Calculate the percentage difference. I then open the audio file of the second capture in an old version of Adobe Audition, although I think something like Audacity will do. Use the Stretch/Pitch function to stretch or shrink the audio file to whatever percentage you've calculated will result in the same number of frames as the trimmed first capture. Save the file.

In Vegas, the saved audio file then automatically replaces the original, but does not change in length. Shorten or lengthen the corrected audio file to line up with the end of scanned film. That should also be the end of the saved audio file, and your audio should now be frame-accurate.

Virtually all the digitizing I do is of home movies, and finding suitable reference points within the film is not always easy. I have better success if I can establish the absolute first and last frame of film. Tail ends of 50-foot reels are often mangled, so a trim with a bit of leader added gives me a clean final frame.

The Elmo gives me reliably consistent speed, provided I let it warm up with about 5 minutes of film. Since I have to speed-correct the audio file anyway, I run the Elmo at 24 fps, even though most home movies were shot at 18 fps, just to save time. In the end, the pitch will be correct. Occasionally, a 400-foot reel of film will still require a frame or two adjustment here and there, but for the most part no other correction is necessary.

Posted by Winbert Hutahaean (Member # 58) on June 20, 2017, 05:56 PM:

Janice, I am not really a tech guy. But isn't there a software that can stretch or squeeze the audio. I found once this software often used by dubber to match with the lips.

The software is very friendly use, you can streetch/squeeze graphically just like you do with picture, by pulling or pushing the audio sound.

I don't know the name of the software.

cheers,

Posted by Will Trenfield (Member # 5321) on June 20, 2017, 06:25 PM:

What was the projector you used, Janice?

Posted by Janice Glesser (Member # 2758) on June 20, 2017, 06:55 PM:

Thanks Ty....I'm using a Workprinter XP also... capturing in HD with Cinecap to a BlackMagic Intensity Pro card. Cinecap doesn't do a good job of converting the HD captures to 24fps...so I have been taking the original file ... flipping the image and converting it to a compressed BlackMagic mjpeg .avi file with VDub. I do this to reduce the size of the file that I transfer to my editing computer. The conversion maybe contributing to the timing issue...don't know. Ty...How do you treat the Workprinter captured files? Do you just put them directly into your editor and change the speed? I could do this in Premiere also...I'll see if that changes anything.

I've used several different projectors to capture sound...but haven't tried my Elmo ST1200 HD yet. For "Fence Buster" I used my Elmo 912S sound viewer to play the film set at 24fps and recorded it to a Zoom H2 digital recorder. It sounded find...but when I put the video and audio on the Premiere timeline the audio was about 45 seconds longer than the video. GRRRGGGH [Eek!]

I've been doing video editing for years...but this was pretty challenging. I probably should have just re-recorded the audio with another projector...but I was stubborn and was going to get it to work. It only took me to 2am to get it in sync.

From what I'm reading...it looks like the audio timing is established by the projector and not the recording device...correct? I'm going to try another projector and see if the sync is better.

@Winbert...yes speed and timing of audio can be adjusted ...but it's usually better to edit the video to avoid distortion in the sound.

Posted by Winbert Hutahaean (Member # 58) on June 20, 2017, 08:18 PM:

Janice, I believe today's audio software does not create the distortion in the sound like a long time ago.

I knew this because I am sometimes DJ ing with the latest Virtual DJ software. A long time ago increasing/decreasing the speed (BPM = Beat Per Minute) will result high or low pitch in the sound. But now, we hardly notice it.

For example for the sound "Yes", if we streetch it longer, instead of creating Yes" sound with low pitch and slow, it is now heard "Yeeeees" but more like people speaking very slow at normal pitch.

So what kind of distortion in the sound you meant?

cheers,

Posted by Janice Glesser (Member # 2758) on June 20, 2017, 09:04 PM:

You can adjust the speed and maintain the audio pitch with Adobe Audition and also in Adobe Premiere (I use both)...but only to a certain degree before it changes the tone of the music or the quality of the voice. Believe me Winbert I've done this over the years many times. However there is a point where the audio especially in a voice-over where the sound of the voice will be altered if the speed is adjusted too much. I can only think of the word distorted because it won't sound natural anymore. In my recent case there was a 45 second difference between the audio and the video and you can not adjust the speed that much and have the voice sound normal.

[ June 20, 2017, 10:08 PM: Message edited by: Janice Glesser ]

Posted by Winbert Hutahaean (Member # 58) on June 21, 2017, 12:01 AM:

Hi Janice,

I am not sure in your case, but I prefer to upload what I did here:

https://youtu.be/V9GNTBoAx74

The original song of Bryan Adams "Straight from the heart" is 190 seconds at normal BPM 83.

This video shows the Virtual DJ 7 which can really produce a distorted sound when we change the speed.

But on the newer deck, Virtual DJ 8, we hardly hear that.

So on the lowest speed BPM 56, the duration becomes longer, i.e 285 seconds or 95 second slower of the original song. And when it is on BPM 112, the duration becomes 143 seconds or 47 seconds faster of the original song.

If you are familiar with Bryan Adams', voice you can hear that no matter on BPM 56 or BPM 83 (normal) or BPM 112, you still recognize his voice, not a Donald Duck or Micky Mouse pitch.

Hope that can explain.

Posted by Ty Reynolds (Member # 5117) on June 21, 2017, 12:22 PM:

Janice - I just used the Cinecap Velocity program for processing the files to the correct fps. Compression doesn't enter into the equation. Neither does pitch correction. If, for example, your film is 10 minutes long, your finished video file will be 10 minutes long, regardless of compression. Once captured and manipulated, your audio file also will be 10 minutes long. Same playback speed, same pitch.

When you capture both the flickering image and the audio directly from the projector, you are creating a visual reference for the audio. By establishing the first and last frame of both the speed-corrected scan and the projector capture, and making them line up, you are basically telling the audio file, "You're supposed to start here and end up there. Make it so."

This will work even if the speed-correction of the scan isn't absolutely perfect. If your 10-minte film becomes a 10 minute and 3 second file, your audio file will end up at 10 minutes and 3 seconds. The difference in pitch in this case is negligible.

All that matters is that the projector performs at a consistent speed.

Posted by Janice Glesser (Member # 2758) on June 21, 2017, 12:33 PM:

Thanks for the demo Winbert. That software is definitely impressive!!! [Smile]

It would be great for modifying the length of background audio and depending on the incremental degree of adjustments it possibly could work for some syncing problems too.

However if the audio isn't captured at a consistent speed to begin with... it can drift in more that one section. Also the process of converting video from one format to another can cause it to go out of sync with the audio. A few milliseconds off at the being of a clip or in another section may not be recognized, but the sync will progressively get worse as the video goes on. This is why just changing the speed of the audio is not always a blanket solution.

I'm going to do some more tests to see what works best with the projectors and software I have. Hopefully I can find a combination that will bring the audio and video together with minimal effort and more predictable results.

If anyone else has experience with frame by frame or realtime telecine with audio...please chime it. I'd love to know what hardware and software you use as well as any conversion processes you do.

Posted by Brian Fretwell (Member # 4302) on June 21, 2017, 04:05 PM:

I've never used any of this software, but if the projector was modified to give a pulse per frame on a second track of the audio - couldn't that be used for sync. Or at least show up any speed variations of the playback.

Posted by Winbert Hutahaean (Member # 58) on June 21, 2017, 06:48 PM:

quote:
However if the audio isn't captured at a consistent speed to begin with...

This means the problem is in your projector. It has high WOW-Flutter (is it the correct term?). Having this, you must take your sound with a projector that has quartz speed or capable to lock the speed, like GS-800 or GS-1200.

I guess if the sound is locked at certain speed, the software can squeeze or streech it at constant level too.

But since there is another variable that might influence the sound, i.e magnetic strip, I don't know what will be the result if the projector has already had a contant speed (locked) but the magnetic strip is warped.

cheers,

Posted by Janice Glesser (Member # 2758) on June 22, 2017, 10:52 PM:

Winbert...I tried the latest version of Adobe Audition CC 2017 and it definitely is better at adjusting the time and pitch. Next time I'll use a more reliable projector to record the sound...but having the better software will help if some tweaking is needed.

[ June 23, 2017, 12:17 AM: Message edited by: Janice Glesser ]

UBB.classic^TM 6.3.1.2