PolyVoice

---

Created to break language barriers, PolyVoice is a prototype/app that me and my 3 other friends created at the Hack OC Hackathon. PolyVoice allows users to translate their voice live into any other language. PolyVoice was voted as the best app built and our team won first place.

Progression

PolyVoice first takes a user's voice as an input, which uses OpenAI's Whisper API to recognize speech. This speech is then translated using DeepL's open API to ensure the translation is conversational and clear.

To collaborate and sync our Python code, we used Visual Studio's live share which we initially had a lot of issues with, but was able to fix. The wix website hyperlink in the title of the section was created as an easier way to present PolyVoice, because none of us had speakers that could easily playback the translated voice.

You can find the github for PolyVoice here and on this website.

Takeaways

The stage I mainly developed was towards beggining where the user would take in mic input and have it be saved to an audio file. This was easily achievable in Python, but issues with Visual Studio live share caused a lot of errors on what computer was taking audio input and output.

I familiarized myself with various Python libraries, including pyaudio and PySimpleGUI. PySimpleGUI allowed the user to choose what language they wanted to translate their voice to from a dropdown box.

Of course, there are still many unaccounted bugs and optimizations due to only having 12 hours to make PolyVoice. We plan to implement more features and fix many issues with our app in the near future.