Octave, a new voice large language model, launched this week. It brings voice AI agents one step closer to their Human equivalents by allowing them to add context, emotion, and nuance to their voices.
Voice AI has grown dramatically in the last six months, and many businesses plan to launch a first layer of AI customer service, before triaging more complex issues for escalation to Human Agents.
The key advantage of AI customer service is that every call is answered instantly at no extra cost. AI agents can be scaled from 1 to 10,000 in a second, allowing any peak in demand to be handled. Crucially, businesses don’t have to pay to have human agents twiddling their thumbs during downtimes because you only pay an AI agent when they are talking to customers.
This saving amounts to hundreds of thousands in cost reductions and arguably provides the customer with a superior experience compared to waiting in a queue to speak with the next available human agent.
My own Neural Voice business, however, acknowledges that not all customers will be content chatting with an AI agent. Therefore, we have introduced the keyword “Human,” which, when used by the customer, automatically escalates the call to a human agent. Importantly, we also provide the option to return to the AI agents if the queue is too long!
Remarkably, our experience thus far indicates that AI can manage virtually any local dialect, as the large language model comprehends the ‘Intent’ of a paragraph rather than the specific words chosen. Thus, even if an individual word is misinterpreted, the intent is typically understood correctly.
Historically, AI voice agents have typically struggled to sound fully human. The speech-to-text tools employed simply articulated the words in the response without incorporating intonation, such as sarcasm, in their meaning interpretation.
However, an AI company named Hume launched Octave this week, the first large language model (LLM) specifically designed for text-to-speech (TTS). Unlike traditional TTS systems that merely read text aloud, Octave is a speech-language model that comprehends context, emotion, and nuance, enabling it to produce lifelike voices with expressive intonation.
As a large language model for speech, Octave doesn’t merely read words—it interprets meaning, rhythm, tone, and structure to produce more natural-sounding voices. For instance, I can detect sarcasm in a response and adjust the tone of he voice accordingly to reflect this. So, if you want a sarcastic customer service agent, these can now be created. It is much more likely that the intonation of an AI agent can be changed to show urgency if a customer makes it clear they have an urgent enquiry or issue.
The broader application involves adjusting the tone of the agent’s voice to align with the tone of the customer engaging with the agent. If the customer is cheerful and playful, the AI voice agent can be adapted to reflect that tone. Conversely, if the customer is particularly serious, it is likely best to respond similarly.
When these tools are combined with the ability to clone individual staff members’ voices or select from thousands of voice options featuring different regional accents, it is easy to see why all businesses, particularly travel companies, should consider implementing AI voice customer service.
Most travel businesses have encouraged customers to utilise online booking channels to reduce costs and enhance conversion tracking. Research has shown, however, that online conversion rates increase by 10% when a prominent telephone number is displayed. Customers feel more assured knowing they can contact the travel business if anything changes or goes wrong.
Many businesses acknowledge this and recognise that phone bookings are often incremental, as these customers frequently cannot reserve adjoining rooms or obtain the assistance they need online. However, many refuse to prominently display a telephone number, knowing it will lead to a significant number of customer service calls, and they do not have enough human agents to manage them.
Hence, the primary advantage of implementing AI customer service could indeed be the capacity to increase bookings by handling additional telephone reservations with the human staff you liberate from these customer service roles.
Whether you’re looking for cost savings or more bookings, you really should understand what Voice AI can do for your business, so if you want some free consultancy, talk to Neural Voice CEO Jeremy Smith by emailing