"Phonetophocus team on OpenAI Whisper Hackathon Hackathon"

Team Idea

You are asked to pronounce several sentences in a foreign language you are learning, like German. Whichever you struggle with, it identifies as having phonological (the "sound-form" of words) space for improvement. For example, if somebody says "Vier" (Four in German), and the system picks up "Fier" or "Fear", the system should tell them, like my ex-girlfriend in real life with this example, that they need a hard "V" sound and a bit of a build up 'FFF' right before. How to do the latter is tricky, maybe with GPT-3. How to recognise phonological sounds, will require a phonological API/module (I have not used any) How to create sample sentences to say, will be ok, as we will use GPT-3 How to check the recognition of the words, will use Whisper and the "translate" (NOT the "transcribe", as that converts to english also). It will be tricky so please feel free to apply to join and take on 1 or 2 of the "Hows" above! Think of it like a pronounciation placement test. In the ideal world, it recommends videos, or uses text-to-speech APIs to pronounce the difference between what you said wrongly, and the words that were said correctly.

Team Members

revisionai

Submission

Team Leader hasn't made a submission yet