Image Processing Reference
Processing Legacy Formats
If you are ingesting audio from some old legacy formats, the players are sometimes
hard to obtain. Vinyl records will give the best frequency response at higher speeds, but
old 78-RPM records are made of a brittle material that imparts more noise.
Because we can alter the sample rates and change the playback time of the audio, it
might be feasible to sample a 78 RPM at 33 RPM in an emergency. The playback speed can
then be increased to the correct value in the digital domain and re-sampled. This also
applies to 45-RPM singles.
A sample rate of 48 KHz at 33 RPM is equivalent to sampling the 78-RPM record at
112 KHz. This might offer the chance to remove more noise from the recording than if it
had been sampled at 48 KHz and 78 RPM. The shape of the grooves and the amount of
movement of a gramophone needle are different for vinyl and 78-RPM disks. You will still
need to fit a cartridge and stylus that is suitable for the kind of disk you are playing, even
if you are sampling at a slower speed.
Low Bit Rate Scenarios
The human eye is more forgiving than the ear. Make sure that the audio still sounds intel-
ligible even if the frame rate goes down. More advanced encoding techniques are being
developed in research labs, some of which are aimed at the mobile phone market and pro-
vide some interesting solutions to lip sync. It may be possible to deploy some of those
techniques. For example, regionalizing the area of interest within the frame might allow
the encoder to work harder on the facial expression and keep the frame rate up so that lip
sync is preserved. It is possible to achieve the same thing by making sure that the presen-
ter stays still and that the background is stationary.
Time to Encode
That's it! We're ready to encode. All the video is ready and the audio is set up. Just select
a codec and kick off the job. In the next chapter we'll tie up the loose ends, talk about
batch-job control and how to initiate the encoder, and then we're done.