lets talk about an html5 webpage that creates a text responsive canvas that makes simple comic style mouth shapes for different sound combinations while play attached audio, the webpage contains one input box above the canvas that accepts a voice mp3 file input, we must use whisper to get and audio transcript, then we must generate a symbol map to feed the canvas mouth shape generator, we should use a predefined phonetic to mouth shape map. then it must play the canvas mouth animation and mp3 at the same time. the page should have minimal loading indications while it processes the mp3 notifying each step. this should be ultra lightweight minimal code. Our sophistication should all go into the mouth shape rendering. Any other steps we can with the audio to improve accuracy we should also do. how is this possible?