I wonder if it is possible to make a simple version to run entirely in the frontend, for short clips?

You can capture video and download it from canvas: https://developers.google.com/web/updates/2016/10/capture-st...

And it looks possible to add separate audio to it: https://stackoverflow.com/questions/39302814/mediastream-cap....

You could also recreate the waveforms and add to the canvas: https://developer.mozilla.org/en-US/docs/Web/API/Web_Audio_A...,

Not widely supported, but you could try to add speech recognition too https://developer.mozilla.org/en-US/docs/Web/API/Web_Speech_...