Post-Processing Audio

Auto-Tune

If you’ve listened to any pop music in the past few years, you’ve most likely heard a number of songs that use the popular post-processing effect called Auto-Tune. Here is a demonstration of how it works:

https://www.youtube.com/embed/9OUgXFZ_WeY

Auto-Tune takes an existing audio file and transforms the bits to make the pitch “perfect.” This is possible because digital audio is represented with bits, which can be easily manipulated mathematically using algorithms.

The alteration of a sound after it has been recorded is called post-processing. Most audio effects (e.g., echo, reverb, phase shifting, backmasking) are added after the recording (a.k.a. post-processing). Digital representations have alleviated many challenges relating to post-processing of audio, which has drastically changed how we view the creation of music.

Transforming Bits During Post-Processing

Post-processing effects are possible on digital images, video, and sounds because we can transform the bits that encode them. These bits are numeric values and therefore can be altered using algorithms (i.e., functions/math operations).

Let’s take a look at how to create an “echo” during post-processing.

  • Pretend that we digitized the sound of a hand clap, represented as the following digital sound wave.

  • How would we represent the sound when it echoes?
  • Since an echo is merely a sound that has bounced off a surface and is re-heard, the sound repeats. Therefore, to simulate an echo, the digitized sound wave should repeat:

  • The same digital sound wave is simply copied.

Note that this would result in the exact same sound repeated, but we could overlap the sounds to make the echo effect more pronounced.

Let’s think about this in another way, using the following “values” that represent a sound wave:

3-1-2-4-4-5

How might an echo look in terms of the values encoded by the digital representation?

3-1-2-4-4-5 3-1-2-4-4-5 3-1-2-4-4-5 3-1-2-4-4-5 3-1-2-4-4-5

Other post-processing effects are more complex, like noise removal, auto-tuning, etc. If you take all the values in a digitally encoded audio file and reversed their order, that would be similar to backmasking. The variety of post-processing strategies is too great to list here and is ever-evolving, but the great thing is you don’t need fancy production software or tools to post-process audio when you have computational thinking and programming skills.