The digitization of analog media is something I’ve developed a particular interest in, and something that’s becoming increasingly vital to anyone looking to preserve and/or archive material stored on older formats.
I’ve done a fair bit of both audio and video transfers (digitization), but with audio I’ve also developed a passion for restoring the audio once its digitized.
The process is a delicate one, and almost always a balancing act between removing perceived noise and preserving the integrity of the desired content.
Restoration Sample
Here’s an example of the process. This was a cassette tape with voice material recorded in 1977 that had a significant amount of broadband noise.
1. Original Source (after digitization)
The first step was to remove a rattle/buzz noise at 6300 Hz (high frequency). Though its subtle to hear, its very obvious in a visual representation of the audio and relatively easy to isolate and remove.
Next, I apply a filter removing low frequency noise below the range of the voices (if your speakers don’t reproduce low frequencies you may not hear a significant difference).
Then I use a tool that learns to distinguish the sounds that are noise from the sounds that are desired content by identifying the unique sound wave patterns of each. Over the course of 2 or 3 passes, I carefully remove as much of the noise as I can while imposing as little impact on the content as possible. For this particular piece I went with a particularly aggressive reduction.
With a good bit of the broadband noise gone we can now hear that there is another noise that occurs intermittently (every 1 to 2 seconds) in the left channel. It sounds almost like someone breathing heavily. If you missed it, listen again to the last sample. We can remove this by isolating and removing just that particular sound whenever it occurred in the recording.
There’s also a slight dropout in the signal at the 3 second mark that I repair.
Leave a comment