New AWS Tools Pull Intelligence from Real-World Media
In its quest to help developers build increasingly intelligent apps, Amazon Web Services (AWS) this week unveiled four new solutions that can turn volumes of unstructured media into data goldmines.
Introduced by AWS CEO Andy Jassy during the Wednesday re:Invent keynote, the four services are Amazon Rekognition Video, Amazon Transcribe, Amazon Translate and Amazon Comprehend. Both Rekognition Video and Comprehend are now generally available, while Transcribe and Translate are still in preview.
Amazon Rekognition Video
Rekognition Video is an object-analysis service that can identify faces, scenes and activities in real-time video, including video stored in the cloud and video that's live-streamed. Its release comes one year after AWS unveiled the original Rekognition service, which identifies objects in static images.
The service uses deep learning to "accurately detect, track, recognize, extract, and moderate thousands of objects, faces, and content from a video," according to an AWS blog post. Rekognition Video also understands context, enabling a capability that Jassy called "person tracking," which lets the service track a person's whereabouts even when they're hidden in a shot.
As its name suggests, Transcribe is a speech-to-text service that converts audio files stored in Amazon S3 (it supports common formats like .WAV and .MP3) into accurate, time-stamped transcripts. It essentially performs the reverse function of Amazon Polly, a text-to-speech engine that AWS also unveiled a year ago.
Transcribe doesn't just do rote word-by-word transcription, according to Jassy. It automatically inserts proper punctuation and formatting, recognizes correct grammar, differentiates between multiple speakers, and allows users to create custom vocabulary libraries.
Another appropriately named offering, this service translates words in text-based formats into any supported language. Currently, the preview supports English, Spanish, French, German, Arabic, Portuguese and Simplified Chinese.
"With the Translate service, organizations and business now have the ability to expand products and services in other regions more easily by allowing consumers to access websites, information, and resources in their preferred language using automated language translations," according to an AWS blog post. "In addition, customers can engage in multiplayer chats, gather information from consumer forums, dive into educational documents, and even obtain reviews about hotels even if those resources are provided in a language they can't readily understand."
Comprehend is an intelligent text-analysis service
that uses natural language processing to pull out information like subjects, key phrases, language and writer sentiment. It supports many sources, including social media posts, articles and reviews.
Comprehend is also able to adapt to colloquialisms and language changes as it's exposed to more text. "Because of the depth and breadth of this training approach, it is able to provide accurate coverage for a wide range of scenarios like analyzing customer feedback, intelligent document search, and automatically organizing content," AWS said in its announcement.
More from AWS re:Invent 2017:
Gladys Rama is the senior site producer for Redmondmag.com, RCPmag.com and MCPmag.com.