Currently, Descript's silence detection feature identifies any segment without speech as "silence" and marks it for removal. This creates a problem for content that includes instrumental music or sound effects without narration.

Use Case:

I'm editing a video where I play bass guitar while not speaking. When I use the "Remove Silence" feature, Descript incorrectly flags these musical sections as silence and removes them, even though they contain important audio content.

Proposed Solution:

Add an audio level threshold to silence detection. Segments should only be considered "silence" if:

There is no speech AND

The audio level is below a user-defined threshold (e.g., -40dB)

This would allow the tool to distinguish between true silence and non-speech audio content like music, ambient sound, or sound effects.