VOCA

Not really used to designing on .svg but after several fixes and support from Copilot, I finally made the logos and banners for the app. It’s no longer speech_coach, welcome to VOCA!

continue the no front-end development streak. For now I’ve just utilized the openAI client with temporary prompt to receive the context and help analyze speech with the exact format. This wasn’t what I expecting from the final app right now, but just a test to see if it’s working. Again, I’ll have to find the logic for more effective coaching. One thing I learned was how Flask couldn’t work and would continuously restart the server if it detected changes in the OpenAI library files.

I added a feature where the user will upload a file that will be saved in /upload and then call an OpenAI’s whisper transcription API to do the task of speech-to-text. Since the app only shows text in mere html, I’ll have to improve its look later on. It’s also important that I have to figure out the coaching logic to evaluate the user’s speech (pace, tone, filler words, audience, or what purpose?). Next step is to focus on AI’s recommendation for users’ speech improvement.

I started foundation with Flask, a basic audio upload form, and several Python code. After several debugging, the project is currently working with a Flask web server with upload endpoint using basic HTML for audio upload. The web is also able to store audio files.