Microsoft is steadily reshaping its Dynamics 365 Contact Center platform to help enterprises enhance their customer service operations with innovations in voice AI.
A common challenge in AI-led customer interactions is a conflict between customer sentiment and an AI voice agent’s response. For example, when a frustrated customer is met with a neutral or overly upbeat response from a voice agent, it can make the interaction feel dismissive and damage the experience.
The tech giant aims to address this disconnect by using high definition (HD) voices from Azure AI Speech.
The HD voices feature is built on neural text-to-speech models and trained on millions of hours of multilingual data to make AI voice agents sound more human, context-aware, and ultimately, empathetic, improving voice-based customer interactions.
Sam Bobo, Senior Product Manager, wrote on Microsoft’s blog:
HD voices can generate speech that closely mimics natural human conversation, including spontaneous pauses and emphasis… [They] can automatically detect emotions in the input text and adjust their speaking tone in real-time to match the sentiment.
HD voices also maintain a consistent voice persona from their neural (and non-HD) counterparts, for more natural-sounding interactions.
This release marks the second new Microsoft voice feature this month, with the vendor also launching Constrained Speech Recognition, a tool designed to introduce structured rules that can increase the accuracy of voice inputs.
Unlike human agents, who naturally use contextual cues, traditional voice recognition systems can struggle to accurately understand what customers say, and especially have trouble comprehending accents, slang and other unexpected wording, or unclear speech.
Constrained Speech Recognition uses structured rules known as “grammars” to help narrow down the words and phrases the customer is likely to use so that automated systems can recognize what they are saying.