Google on Wednesday unveiled a new machine-learning platform for developers.
Available now in limited preview, Cloud Machine Learning provides access to the same technology that powers Google Now, Google Photos, and Google Search voice recognition. It's also used in Google Translate and Inbox's new Smart Replies, and promises better training performance and increased accuracy compared to other large-scale deep learning systems.
Developers can sign up now to build models from their own data, or tackle a pre-trained model, like the new Cloud Speech API. The latter, Google boasted, demonstrates speech-to-text conversion in more than 80 languages, "with unparalleled accuracy, especially in noisy environments."
The free API, which takes aim at voice-recognition company Nuance and Apple's virtual assistant Siri, is open in limited preview to third-party developers; pricing will be introduced later, according to TechCrunch.
"Cloud Machine Learning will take machine learning mainstream, giving data scientists and developers a way to build a new class of intelligent applications," Fausto Ibarra, director of product management, wrote in a blog post.
Portable, fully managed, and integrated with other platform products, the machine-learning service lets developers build models of any size that work on any type of data.
As The Verge points out, Google isn't the first to introduce a machine-learning platform: Microsoft Azure and Amazon Web Services launched their own in 2014 and 2015, respectively.
Google in November open-sourced its TensorFlow machine-learning system, allowing the community to build upon the existing program framework, provide feedback, and contribute to the source code. Now, folks can use TensorFlow Service other another of Google's open-source projects, Kubernetes.
Wednesday's announcements were made during NEXT 2016, Google's Cloud Platform Global User Conference in San Francisco.