Audio bitrate is the data rate of the sound, separate from the picture. Stereo AAC at 128 kbps is the usual default and the figure this site budgets with; speech starts to sound thin below roughly 96 kbps, and music below 128 kbps.
On a long video the audio is a rounding error. On a short one it is not, and this is where people get caught. A 15-second clip compressed to fit a 2 MB limit has about 1,090 kbps to spend in total. Take 128 kbps for stereo audio and you have handed over nearly 12% of the budget before the first frame. Drop to mono at 64 kbps and you get most of it back.
Two practical consequences:
- On a tight, short target, check the audio first. Halving it is free in perceptual terms for speech, and it buys picture quality directly.
- On anything over a minute, leave it alone. Saving 64 kbps across three minutes recovers about 1.4 MB. It is not where your file size is.
There is also the case of no audio at all. Screen recordings without narration, silent B-roll and GIF-style loops often carry a full stereo track of nothing. Stripping it is the cheapest size reduction available and costs exactly nothing, and it is worth checking before you touch a quality slider.