awexa (´・ω・｀)™️

he...hewwo wowwd???? (´・ω・｀)

using expert minimalist artisanal japanese construction techniques (origami), we've built an eco-friendly smart assistant from renewable materials (used coffee cup).

Inspiration

sleep deprivation

What it does

awexa (´・ω・｀)™️ is your one-stop-shop fow all your voice-to-text-to-voice needs. you speak, awexa (´・ω・｀)™️ listens, and then awexa (´・ω・｀)™️ speaks back to you. its wike hawing your own personal wobot assistant, but without the wobot body.

How we built it

1 x jbl clip 3 speaker (won at lucky draw from hack n roll 2021)
1 x half-washed food clique kopi c kosong cup (with coffeestains)
1 x paper (from table)
1 x double sided tape (acquired from residential college 4 level 6 lounge)

we created an internal RESTful API with FastAPI to do the user registration and voice transcription functions.

the voice input (and output function) is handled by our voice_input POST endpoint, which first uses the OpenAI Whisper model on Replicate to do transcription, then forwards it to GPT-3 to generate a response. text-to-speech audio output is generated using the Google Text-to-Speech API.

the front-end was created with reactjs.