Inspiration
I had to start with an actual ML deployment, this couldn't have been a better time, hence I tried to deploy a transformers model( word2vec2 precisely ) using heroku and flask
What it does
It just converts speech to text. There is a chonky piece of model that converts the audio to vectors, and then to text, in English
How we built it
I first experimented with the notebook, and got the model working, then I made a standard Flask app that serves a single page for the purpose
Challenges we ran into
I was a beginner at flask, today I deploy my first flask application. So there were some problems here and there. For now I am struggling with deployment, you can see the deployed product once it is done in the link given below
Accomplishments that we're proud of
- Got a flask app to work for the first time trying it
- Worked with a Pytorch model
What we learned
- Handling audio/wav files
- Working with PyTorch models
What's next for speech2text
- Deploying it properly
- Using it as an api for other applications
Log in or sign up for Devpost to join the conversation.