Software to Read Variable Density Optical Soundtrack

[jersy]

Hello All,

Trying to figure out a way to pull this off, hoping you can help.

I found approximately 10 seconds of shot and processed 16mm film. This image area is blank, but there is a variable density optical soundtrack on the film. I'm hoping to scan the individual frames and see what is coming from the soundtrack - hopefully to get some idea as to what the content of the reel might have been. Worst case scenario is that there is nothing there, but I actually learn something.

Can anyone offer any suggestions as to what software I could use to accomplish this task? I've looked for the past week or so, and cant quite find something that will work.

Thanks for any info.

[jagabo]

A few of us played around with that a few months ago.
https://forum.videohelp.com/threads/334883-Scanned-16mm-Optical-Soundtrack-Direct-to-Digital-Audio

[jersy]

A few of us played around with that a few months ago.
https://forum.videohelp.com/threads/334883-Scanned-16mm-Optical-Soundtrack-Direct-to-Digital-Audio

Thanks. I saw that earlier, but since it focused on variable-area sound, I wasn't sure it would apply. I've grabbed the NAudio software mentioned in the thread, but havent been infront of a Windows machine yet to try it out.

[jagabo]

It's basically the same thing isn't it? Just sum up now much light passes through a certain area. Do you have a scanned sample you can post?

[jersy]

It's basically the same thing isn't it? Just sum up now much light passes through a certain area. Do you have a scanned sample you can post?

Sure. Guess I'm not clear about the calculations yet.

[jagabo]

I'm pretty sure it works like this: light is projected through a narrow horizontal slit and passes through the film. If falls on a photoamplifier that converts the light into a voltage. The more light the receptor sees the higher the voltage. So you can directly convert the brightness to a waveform.

That's not really enough to tell what the sound might be. But if you take one column of pixels (or average across rows) and export as raw 8 bit grayscale data you can import it into Audacity as raw unsigned 8 bit audio samples. I imported it assuming 8 KHz but it just sounds like a muffle with a pop. Attached is a WAV file with the segment repeated four times. You'll probably need 10 times more data, and in an area that's more "interesting", to get an idea what the sound is.

[jersy]

I'm pretty sure it works like this: light is projected through a narrow horizontal slit and passes through the film. If falls on a photoamplifier that converts the light into a voltage. The more light the receptor sees the higher the voltage. So you can directly convert the brightness to a waveform.

That's not really enough to tell what the sound might be. But if you take one column of pixels (or average across rows) and export as raw 8 bit grayscale data you can import it into /snip/ as raw unsigned 8 bit audio samples. I imported it assuming 8 KHz but it just sounds like a muffle with a pop. Attached is a WAV file with the segment repeated four times. You'll probably need 10 times more data, and in an area that's more "interesting", to get an idea what the sound is.

Thanks for that. It's a start. I'll look closer at Audacity tonight. It's now to see if Audacity will take multiple frame scans and stich them together, or if I'll have to do each one individually, and then change the duration of each result to 1/24 of a second. (if I'm understanding this all correctly).

/Edit/ Or perhaps just one long image stitched in Photoshop or something similar.

[jagabo]

Audacity is a full featured audio editor so you can indeed append and manipulate segments. I would scan in as large segments as you can (14 inch?) with as much resolution as you can. Work in an editor that supports 16 bits per primary and export as raw 16 bit grayscale. Import that into Audacity.

[aedipuss]

for 10 seconds you should have about 6 feet of film. you could do a foot at a time(40 frames), but for timing purposes it may be easier to work with just 24 frames at a time.

i doubt there's anything worth recovering from developed film an editor has thrown out. most likely it's lead-in and all you are going to get is the cam operator saying "speed".