Overview

Introducing the concept of improving ASR(Automatic Speech Recognition) system accuracy and efficiency through data type hints.

Adding Data Type Hints for Voice Bot

• When collecting specific information (e.g., mobile number), provide a data type hint to the ASR system.

  • Pass a hint like 'numeric' to bias the ASR system towards recognizing numeric values.

  • Example: Ask customers for their mobile number using the 'numeric' hint to improve extraction accuracy.

Utilizing Language and Entity Values for Keyword Queries

• Use language and entity values to enhance the voice bot's understanding of keyword queries.

  • Create distinct entity values for different color types (e.g., Debit Cards, Visa Cards, Master Cards).

  • Example: Configure entity values to correctly identify a user's intention to block a debit card when they mention "Visa card.

Biasing ASR System for Specific Keywords

• Bias the ASR system to recognize specific keywords or phrases.

  • Pass relevant keywords to the ASR system to improve recognition accuracy.

  • Example: Pass "previous month" or "current month" as keywords to facilitate accurate extraction of the desired time frame.