Posted On: Nov 20, 2019

Amazon Transcribe is an automatic speech recognition (ASR) service that makes it easy for you to add a speech-to-text capability to your applications. Up until now, Amazon Transcribe returned transcriptions with the highest confidence score only. Starting today, you can specify that Amazon Transcribe return additional transcriptions as well. You can use these alternative transcriptions to see different interpretations of the transcribed audio, giving you added optionality and flexibility to select the most suitable text results for your application. For example, in media asset management (MAM) applications that enable users to review and splice together subtitles, users can now select from up to ten alternative transcriptions. Alternative transcripts are available across all languages supported by Amazon Transcribe.

Amazon Transcribe enables organizations to enhance the analysis, accessibility, discoverability of their audio and video content. The service is useful across a breadth of use cases. Customer contact centers can convert call recordings into text and analyze the data for actionable intelligence. Media content producers can automate subtitling workflows for greater engagement and accessibility. Law firms and law enforcement agencies that have recorded meetings or court proceedings can make those archives searchable by transcribing them into useful text that’s demarcated by time stamps. Also, marketers and advertisers can enhance content discovery and display targeted advertising based on the extracted metadata from video files (such as showing a sports shoe ad in the middle of a football match vs. a cooking show). 

The support for Alternative Transcriptions is available in all the AWS Regions where Amazon Transcribe is available at no additional cost. Try it through the Amazon Transcribe console, or use the Command Line Interface (CLI) and AWS SDKs. For more information, visit the documentation page.