Contact Center Technology & Voice Automation
Contact centers rely on a variety of technologies to provide efficient and valuable customer support. These technologies include networks in place that allow receival and transmittal of many types of communication such as calls, emails, social media, and live web-based chat. Additionally, contact centers today tap into technologies including, but not limited to AI, chatbots, automation, and natural language processing.
Incorporating Voice Technology
Technology is always improving, and many contact centers frequently update their tools, technologies, and services. Since most contact centers already have a foundational technology infrastructure in place, it is important that new tools and technologies are able to integrate with existing services.
Text-based support channels involving human agents are relatively easy to extend with chatbot functionality as they are designed to process text-based communication. Yet most contact center interactions occur on voice-based channels (such as SIP Trunk, PSTN/cellular, WebRTC, etc.) However, as voice-based inquiries are relatively costly and inefficient to handle for human agents, it is desirable to use virtual agents to resolve simple requests. One solution is to use a tool such as Cognigy Voice Gateway, which allows contact centers to integrate voice bots into their processes.
The gateway is a flexible solution that can connect existing voice networks with cognitive services, including conversational AI frameworks, speech-to-text (STT) engines, and text-to-speech (TTS) engines. It also provides advanced call management functions such as call disconnect, call transfer, and call recording.
When a customer connects via a voice-based channel, Voice Gateway provides a link for information to flow between a chatbot service and the customer as shown in the figure below
Voice Gateway consists of two primary components – the voice engagement channel and cognitive services. The voice engagement channel is responsible for interfacing with voice-based channels. Its capabilities include media handling, security, translation, SIP interoperability, high availably, and scalability. The underlying session border controler (SBC) manages the information flow between Voice Gateway and outside services such as bot frameworks and STT/TSS engines. Voice Gateway can easily integrate with existing frameworks and services, as shown in the figure below.
Contact centers that use bots rely on many speech-related technologies such as continuous automatic speech recognition (ASR), speech synthesis markup language (SSML), and barge-in. ASR enables bots to recognize when a person is speaking. SSML allows bots to deliver customized audio responses and can be used in conjunction with TTS engines to provide details on pauses, text that should be censored, and audio formatting for unique text such as dates and abbreviations. Barge-in refers to when a user interrupts a bot when it is speaking; it is important to consider how a bot will respond to barge-in.
The Voice Gateway utilizes all of the above speech features, in addition to others such as language configuration and TTS caching. Continuous ASR enables Voice Gateway to collect speech from a user; it can detect silences and concatenates text segments output from STT engines into a single message to the Conversational AI to ensure that no segments are cut off. When a user barges-in on a conversation, the AI can be programmed to either ignore the interruption or immediately stop responding and process the new speech from the user.
Conversation Initiation and Call Control
The ability to customize and optimize communication with customers via conversation initiation and call control is invaluable for contact center operations. Conversation initiation includes activities such as SIP messaging, establishing a connection, and welcome messages between a bot and a user.
A good conversation initiation process is critical because it is the first step in customer support – without a good connection or the appropriate initial messages, customers may be unable to get the support they need. Similarly, call control abilities such as call transfer are essential features because they allow calls to be managed based on the customer’s needs. For example, if the Conversational AI is unable to resolve a customer query, call transfer enables the customer to be seamlessly transferred to a human agent.
An Enterprise Voice Gateway solution hence needs to provide features for both conversation initiation and call control. When a user begins a conversation, the Voice Gateway sends an initial activity message to the Conversational AI; the content of this message can be adjusted, or the message can be disabled. Controls are also in place for when the Voice Gateway connects a virtual agent to a user and when/if it sends an initial message to the AI and welcome message to the user.
In contact center setups, it is common for IDs associated with calls to be required by Conversational AI applications and sent to the AI in a SIP header. Furthermore, Voice Gateway can be configured to extract values from the SIP INVITE message and include them in the initial message to the AI. Call transfer and disconnect features are also included, which allow the virtual agent to easily transfer or disconnect the call at any point during a conversation.