Posted On: Jan 31, 2022

Amazon Lex is a service for building conversational interfaces into any application using voice and text. We are excited to announce multiple transcripts and confidence scores support for speech input. The transcripts provide alternate interpretations of the user's speech input. Each transcript is associated with a confidence score that indicates the likelihood of a match. The transcripts along with the confidence scores can be used to enhance the conversation design.

You can use the confidence score to implement an efficient dialog. For example, consider a banking bot that requests information from a user (“What’s the name on the account?”) and the user provides a response (“Michelle”). If the transcription confidence score indicates moderate confidence, then the bot can be configured to clarify (“Did you mean Michelle?”). Alternately, if the score is low then the dialog can be designed to repeat the question (“Can you please say that again?”). Similarly, you can use the transcript along with business knowledge to improve your understanding of the user’s response. For example, in case of an inventory status bot you can apply business logic to process the multiple transcripts (“Scissors”, “Cesar’s”, “Seizures”) and identify the appropriate interpretation (“Scissors”). You now have better insights to manage the conversation flow and create more effective conversations.

Multiple transcripts and confidence scores are available in all AWS Regions where Amazon Lex operates. To learn more, visit the Amazon Lex documentation page.