Google Cloud Platform Releases Two New Cloud Machine Learning Products

1024 496 The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)

This post is an excerpt from the July 22, 2016 edition of the This Week in Machine Learning & AI podcast. You can listen or subscribe to the podcast below.

The Google Cloud Platform team made news this week, with the announcement of the public beta of two new Cloud Machine Learning products: the Cloud Natural Language and Cloud Speech APIs.

The Cloud Natural Language API is based on NLP work developed by Google Research including the recently open-sourced SyntaxNet and Parsey McParseface, which analyzes the grammatical structure of text with high accuracy, speed and scale. The new API offers three features:

  • Sentiment Analysis for understanding the overall sentiment of a block of text.
  • Entity Recognition for identifying the major entities referred to in the text, for example people, places, and events.
  • Syntax Analysis for identifying the parts of speech in a block of text, and producing a parse tree for each sentence.

The API currently works with English, Spanish and Japanese input.

Google’s also posted a blog that walks through using the new API to analyze text from Harry Potter and The New York Times.

The other API, the Cloud Speech API, offers text-to-speech conversion with support for over 80 languages. It’s based on the same speech core recognition technology used in Google Search and Google Now. The beta adds new features that weren’t in the alpha, including:

  • Word hints, which allows custom words or phrases to be added to API calls to improve recognition. App-specific commands and names are two examples where this might be used.
  • Asynchronous support, which simplifies development for some applications.

The Cloud Natural Language API is free for up to 5,000 uses per month, and costs $1 per use thereafter with a 50% discount while in beta. The Cloud Speech API is free for up to 60 minutes per month, and then costs 6 tenths of cent per minute thereafter.

Leave a Reply

Your email address will not be published.