The video is nearly all of the file size
In a typical recording the picture accounts for the overwhelming majority of the file and the audio for a small fraction of it. A screen recording of a long meeting can be several gigabytes of which the speech is a few dozen megabytes, because moving images need vastly more data than a voice does.
Since we read only the audio and ignore the frames entirely, extracting the audio track before uploading costs you nothing in accuracy and can shrink the upload by a factor of fifty or more. Any basic video tool will export the audio as a separate file, and that file is what you upload.
When to extract and when not to bother
If the video uploads comfortably, upload it as it is. There is no benefit to extracting audio from a file that already fits, and every extra conversion step is another chance to introduce a problem or pick the wrong track.
Extract when the upload is failing, stalling repeatedly on a slow connection, or sitting above the size limit for your plan. Choose a lossless or high quality export when your tool offers the choice, since there is no reason to compress audio that is already small compared to the video you just discarded.
Frequently asked questions
What is the largest file I can upload?
Pro accepts up to two gigabytes and five hours in a single file. The free first file is capped at fifty megabytes and ten minutes, and a free account raises that to thirty minutes and one hundred megabytes.
Does extracting the audio hurt accuracy?
Not if you export at reasonable quality. You are discarding the picture, which we never look at anyway. Only re-compressing the audio heavily during export would make any difference to the result.
My upload keeps failing partway through. Why?
Usually an unstable connection rather than the file. Extracting the audio makes the upload dramatically smaller and therefore much more likely to complete on a connection that keeps dropping.