Conversational AI 101: NLU and IVR for Beginners
Artificial Intelligence is what makes automating voice conversations possible. Companies such as Google and Amazon are continually bringing new iterations of conversational AI technology into the mainstream and integrating different applications of machine learning into our everyday lives.
Asking Alexa to play your favorite podcast and have her quickly play it for you is made possible by the same type of technology that allows contact centers across the world to automate voice conversations.
Conversational AI is making the contact center and customer service industries better than ever.
Having a good understanding of how this happens will help those in the industry to best leverage that technology for their companies.
The ability of conversational AI technology to detect and understand speech are two of the biggest challenges faced by the Intelligent Virtual Agent (IVA).
Understanding the difference between detecting and understanding speech will shed light on how advanced and complex the system needs to be in order to successfully automate your routine through complex phone calls. The development and management of these systems is also a great reason to put your IVA application into the hands of an experienced and knowledgeable provider.
Speech Detection vs. Recognition: Understanding What the Customer is Saying
Speech detection determines when the customer starts and stops talking.
This can be extremely difficult for systems due to several factors including background noise, cross talk, low quality audio connections, and other factors. In order to prepare for these issues, IVA systems leverage machine learning, which analyzes millions of utterances (a spoken word, statement, or sound) and conversations across millions of calls to “train” on when utterances start and stop. This enables modern systems to be extremely accurate when distinguishing between customer speech and background or auxiliary noise. Then, adding in experienced software engineers to fine tune the system results in an IVA that is extremely accurate in speech detection.
Speech recognition is identifying the words that a customer says.
Speech recognition is the second area for an IVA to distinguish itself in customer service applications. Words spoken by customers are broken down into individual sounds or “phenomes” to be analyzed by systems on the acoustic and syntactic characteristics of those sounds.
Language modeling technology predicts the sequence of words that were most likely spoken. The system then puts this sequence into text form and processes the input of that text. This is where the NLU comes in!
Natural Language Understanding (NLU): The Next Level of Speech Recognition
One way that IVA “levels up” from the typical IVR is that an IVA will utilize natural language understanding, or NLU.
This is the ability to extract meaning and intent from the sounds and words that have been recognized.
NLU is capable of listening for and understanding context, as well as deciphering bad grammar, mispronounced words, and distinguishing between similar sounding words.
The ability for a customer to interact with a computer in a way that is virtually indistinguishable from another human is a gold standard for the modern IVA.
NLU utilizes sample data, or “training utterances” to tell the system what it should be listening for. The sample data used to build the IVA is how NLU technology accurately interprets what a customer is saying, and what their goal is in the conversation. Then, talented software engineers identify correct and incorrect examples in the NLU to update its models and improve accuracy.
IVA and the utilization of NLU is a game changer when it comes to the experience customers will have when interacting with the contact center.
An IVA is not only capable of having a seamless and complete conversation with a customer, but they take pressure off human agents in contact centers.
Utilization of an IVA has numerous advantages, even being proven to handle some conversations better than their human counterparts.
Understanding how IVA and NLU operate is a step towards creating an excellent customer experience and simultaneously enhancing efficiency for the contact center.
Since 2002, SmartAction has helped 100+ industry-leading brands streamline their contact centers and take their customer experience to the next level through AI voice, text, and chat.
Discover why we’re the top-rated Virtual Customer Assistant solution on Gartner Peer Insights and have been named “The Leader in AI-enhanced Self-Service,” by Frost & Sullivan.
GET IN TOUCH
See How We Can Help You Exceed your CX Goals.
How Virtual Agents Help the Live Agent Workforce
How exactly does the real–life relationship between virtual and live agents play out? Check out this blog post to find out.
Masterclass: Conversational AI for Contact Centers
Learn how AI experts can tune your customer grammars and intents for absolute accuracy beyond your home device or phone speech-to-text.
Debunking 5 Conversational AI Myths
We expose the top 5 conversational AI myths for what they are and set those up for success who are forging a trail towards automating their contact centers.