Most recordings are not saving the original audio as-is but perform a compression. While compression is often necessary for practical reasons (like saving storage space or reducing bandwidth), it’s a delicate balance. Too much compression can strip away the authenticity of the original call.

When you compress a recording, you’re essentially reducing the amount of data it contains. While compression algorithms are designed to remove redundant information that may not be easily perceivable to the human ear, they can sometimes go too far.

Imagine you’re listening to a song, and certain subtle nuances or details in the music are lost during compression. It could be the delicate strumming of a guitar string or the nuances of a singer’s voice. These small details contribute to the authenticity and richness of the original recording.

Highly compressed recordings may also suffer from what’s known as “compression artifacts.” These are unwanted changes or distortions introduced during the compression process. Think of it like a photocopy of a photocopy—each generation loses a bit of fidelity. The compression artifacts can be even more severe when the original audio was modified by an aggressive AI algorithm – for example an aggressive noise cancelling software. The reason is that an aggressive noise cancelling software while attempting to remove the noise also modifies the original voice in a way that can be significantly different from the original human voice. The human ear might not be disturbed when listening to this artificial modification human voice but a compression algorithm might find it disturbing and difficult to handle and the result might be an increase in the compression artifacts.

To summarize, whenever you introduce a new audio filtering, like an aggressive noise cancellation software, you should listen to your recordings and based on their quality you might want to modify the aggressiveness of the audio filtering and/or the compression rate of the recordings.