Richardson Leao Posted June 18, 2007 Share Posted June 18, 2007 Dear all, I am finalising a little system I built which records the sound and the pilot tone of my kinor 16. the pilot tone gives the motor frequency (fps x 2). So, once I have the pilot frequency and the sound, I can resample the sound to match exactly the telecine frequency, for example, my camera was running at 24fps +-2% the telecine was done at pal or whatever (23.??? fps), so I just resample the sound file as a factor of pilot fps/telecine fps. My question is, before the computers, what would happen with pilot tones? I mean, I know nagras recorded them also, so, how did people do in the past to match the film speed with the sound speed? Cheers richardson Link to comment Share on other sites More sharing options...
Don Brown Posted June 18, 2007 Share Posted June 18, 2007 Dear all, I am finalising a little system I built which records the sound and the pilot tone of my kinor 16. the pilot tone gives the motor frequency (fps x 2). So, once I have the pilot frequency and the sound, I can resample the sound to match exactly the telecine frequency, for example, my camera was running at 24fps +-2% the telecine was done at pal or whatever (23.??? fps), so I just resample the sound file as a factor of pilot fps/telecine fps. My question is, before the computers, what would happen with pilot tones? I mean, I know nagras recorded them also, so, how did people do in the past to match the film speed with the sound speed? Cheers richardson Hi Richardson In the early days the Sound Dept on a film would control the camera the optical sound recorder and camera had interlock sync motors and the Sound Camera operator would shout sound speed when they where running, this carried on with magnetic film. When the 1/4" tape came in there where lots of ways to hold sync, the BBC had a 16/35mm Magnetic tape bay that the motor was run from a amp that was feed from the pilotone recorded on the 1/4". Nagra had a unit called an SLO which was a comparator with an Ocillascope tube that you would compare the incoming sync from the Nagra and the reference frequency ie mains or crystal this would be seen as a circle on the screen. The later Nagra's have their own sync lock board built in which comes into operation when the machine is in playback mode. Regards Don Link to comment Share on other sites More sharing options...
Nick Mulder Posted June 18, 2007 Share Posted June 18, 2007 I am finalising a little system I built which records the sound and the pilot tone of my kinor 16. the pilot tone gives the motor frequency (fps x 2). So, once I have the pilot frequency and the sound, I can resample the sound to match exactly the telecine frequency, for example, my camera was running at 24fps +-2% the telecine was done at pal or whatever (23.??? fps), so I just resample the sound file as a factor of pilot fps/telecine fps. Gidday, I once scoped my Bolex EL to find out the actual running fps of the different (non xtal) speeds ... Found most were a little slow a factor of around %1 or 2 - the 12fps setting being he odd one out was slightly fast ... I could trust these figures and do the same process as you mention - I'm not sure though, is your process dynamic ? ie. does it cater for fluctuations in speed throughout the shot ? If so, then how are you translating the tone/pitch information into the language your time-stretch/compression facility uses ? MAX/MSP ? differential equations! If it weren't dynamic how often do you 'take readings' - once and never again ? per take ? Interested ;) Link to comment Share on other sites More sharing options...
Richardson Leao Posted June 18, 2007 Author Share Posted June 18, 2007 hey Nick, the pilot tone is recorded together to the sound (same sample rate... i know it's exageration but it's easy to deal with the same number of samples and i don't think you can run out of disk space until after you shoot 300 mags...), so i do a windowed fft (of 200ms length or so) of pilot tone data. I was also thinking about using the pilot tone to denoise the sound but after discovering a sound proof foam (strasonic), I could take my kinor to the opera. Also Don, that;s a nice piece of info... i like the shouting fps... Gidday, I once scoped my Bolex EL to find out the actual running fps of the different (non xtal) speeds ... Found most were a little slow a factor of around %1 or 2 - the 12fps setting being he odd one out was slightly fast ... I could trust these figures and do the same process as you mention - I'm not sure though, is your process dynamic ? ie. does it cater for fluctuations in speed throughout the shot ? If so, then how are you translating the tone/pitch information into the language your time-stretch/compression facility uses ? MAX/MSP ? differential equations! If it weren't dynamic how often do you 'take readings' - once and never again ? per take ? Interested ;) Link to comment Share on other sites More sharing options...
Nick Mulder Posted June 18, 2007 Share Posted June 18, 2007 200ms ... okidoki - fourier, I knew bloody math would turn up - and the opera to boot! so you ae making your own programs ? or using ? now how would the pilot tone reduce the noise ? Aside from simply helping with with tonal variancies that would ride in 'harmony' with the speed/pitch of the motor ? How would you teach sound soap or ??? to learn/read this info ? Speaking of Opera, you'll end up with bum notes if you pitch that - so I'm guessing you are using stretch/compression instead, you are using a shorter um whats the term 'time-base' (?) than 200ms for that huh ? To avoid chop-chop choppy audio ? Link to comment Share on other sites More sharing options...
Richardson Leao Posted June 19, 2007 Author Share Posted June 19, 2007 I am still starting with the whole thing, but i might also use another freq/time transform with the pilot signal (maybe wavelets?), at the moment, i have to versions for the program, one in linux that does not support my 24bit sound card (in C) and the other one in labview in windows that support my sound card. i am trying to get the whole thing self powered (valved preamps, mixer, soound card) in an aluminium case and have it on the go (4 tracks etc). The compression/stretching i am doing in matlab, so i could try wavelet ridges as it would give a sort of curve (like compresion ratio vs time). The thing with the pilot tone and noise, if you use a wavelet transform to denoise the signal, you end up decomposing the signal as a series of harmonics, i would then fit the pilot waveform to the decomposed sound, find where it matches and eliminate the matching part independent on phase. But as I said, the blimp is working well enough to not require postprocessing of the sound. Once I have it properly tested, i'll make my progs publicly available. 200ms ... okidoki - fourier, I knew bloody math would turn up - and the opera to boot! so you ae making your own programs ? or using ? now how would the pilot tone reduce the noise ? Aside from simply helping with with tonal variancies that would ride in 'harmony' with the speed/pitch of the motor ? How would you teach sound soap or ??? to learn/read this info ? Speaking of Opera, you'll end up with bum notes if you pitch that - so I'm guessing you are using stretch/compression instead, you are using a shorter um whats the term 'time-base' (?) than 200ms for that huh ? To avoid chop-chop choppy audio ? Link to comment Share on other sites More sharing options...
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now