I treated myself to thinking about the single channel variable density audio on a neg track and wrote an algorithm to try and restore the audio.
I did not re-scan the film but started with one of the single distorted channels from the push-pull example. The was a wee bit of a challenge as I had to try and correct for the density curve of the negative at the same time as correcting for the grain noise reduction mechanism of the variable shutter.
None of these files are bandwidth limited or de-emphasized. I included the wave files so you could listen to a cleaner source. Also. Forgive the odd sample rate.
The digital background noise level is higher than needs be as I went through a reduction application step used for other soundtracks that is not required (I don’t like to change too much software at once.
In theory, if I had a high-level tone test or dual tone test I could optimize the parameters in the correction algorithm, but for now I just guessed. There are techniques to improve the audio without reference material but they involve writing a whole bunch of optimization software.
