Creating .mp3 audio tracks would not be super difficult technically but I feel like this is better handled by the person who is working with the individual files they're working with who knows exactly what that individual tool requires and can create it specifically on demand using their standard tooling.

Moreover I would expect captioning services to *want* the video to provide context and wouldn't consider a divorced-audio-only-using service to be acceptable for eg content translation purposes.

I'm going to provisionally close this as "declined". If there's a very strong desire to reopen this, please provide more detail about the individual tools used and examples of how you're extracting audio and why it's problematic to do it using your own tooling.