What it does
Dictate a list of ingredients you are interested in using, and Speech to Recipes will show you relevant recipes from the McCormick recipes API.
How we built it
We pull data from the McCormick API and store it in an elasticsearch instance. From elasticsearch, we can query recipes based upon their ingredients.
On the front end, we interface with the Web Audio API to record the user's recipe query. We use React, Redux, and Semantic UI to dynamically render recipes in a sleek card-based layout.
With Node.js back end, we process user's voice data by
- Sending voice data to bluemix's speech to text API
- Parsing out ingredients from user's query
- Search elasticsearch for recipes containing the user's requested ingredients
Challenges we ran into
Wanted to be able to query recipes based upon ingredients, but the McCormick API did not directly support this. So we indexed the data with elastic search to more easily perform our queries.
Accomplishments that we're proud of
Fully functional product, deployed to https://www.spicytalk.org.
What we learned
Working with bluemix and McCormick API. Using Web Audio API for recording audio.
What's next for S2R
Improve robustness of NLP for parsing out ingredients in user queries.