The goal of this meetup was to setup a specch recognizer to answer queries from the InstaKart dataset. We divide the problem into 3 parts, of which we have achieved only the first two.

  1. Find a speech recognition API and make it work.
  2. Query on a given product information. More specifically, recognize a product name and look for similar products in the dataset and return the location of the product.
  3. Make the products returned in part 2 relevant. Beef up the speech recognition to get semantic meaning.

The first part was more complicated than expected. Eventually we made it work, the idea is to download the Python Speech Recognition API and all its required dependencies, in particular we need pyaudio to get audio from a microphone.

For the second part we download the InstaKart data and create methods that lookup for a key word among all the products. See more details here.

⤧  Previous post Feature Speaker June 2017 ⤧  Next post Pictures from last week's featured talk.