Voice Search Sees Business Boost

Another strong use case for enterprise voice search is in financial services, where the technology can mine both sides of conversations to start searching for the annuities, mutual funds, or credit cards that callers already have and feed that information to advisers so they can offer only products or services that make sense for particular account holders.

USAA uses voice search technology to recognize utterances within customer conversations with agents in real time. If a life-changing event like a new child is mentioned during the conversation, the technology can call up information about life insurance for agents to offer.

Additionally, agents can use voice search to call up specific data, such as the last five transactions or payments made, to provide context for current requests.

Voice Technology Advances

While consumer interest in voice search technology is the major contributor to growth on the enterprise side, technology advances have also played a significant role. Speech systems that enable voice search have advanced tremendously over the past few years, according to Evan Macmillan, CEO and cofounder of Gridspace, a provider of technology for real-time call scanning. Five years ago, it took 11 hours to transcribe one hour of conversation. Now those transcriptions occur in real time, he says.

“You need to be able to understand speech in real time,” Macmillan says.

Natural language understanding is the linchpin underlying the efficiency of voice search, experts agree.

Without natural language understanding, agents need to know exactly what to ask with some precision, Snell explains. Natural language understanding can understand what agents are asking, even if the wording isn’t exact, then respond back to them.

“Everything has changed in the last five or six years,” says Alexander Lazutin, founder of ICarta Technologies, a software development firm for mobile and web apps, chatbots, and voice services. “It all started with Google Voice. Now [speech recognition] has very low latency at a very affordable price. Speech recognition is becoming so popular with the public that it’s really needed for enterprises.”

Voice search and speech recognition have also advanced over the past couple of years thanks to continuing development around the Generative Pre-trained Transformer 3 (GPT-3) and Bidirectional Encoder Representations from Transformers (BERT), adds Steve Levine, chief marketing officer at Cortical.io, a provider of natural language understanding solutions.

GPT-3 is the third generation of OpenAI’s auto-regressive language model that uses deep learning to produce human-like text. OpenAI, an artificial intelligence research laboratory, introduced GPT-3 in May.

According to OpenAI, GPT-3’s full version has a capacity of 175 billion machine learning parameters.

Microsoft also developed a language model that many companies use today. Its Turing NLG, launched in February, claims a capacity of 17 billion parameters.

The business benefits from voice search also extend to and come from voicebots, which are gaining acceptance by both consumers and workers alike.

More than three-quarters (76 percent) of organizations responding to a recent Capgemini survey reported measurable benefits from voice and chat assistants, and 58 percent said those benefits met or exceeded expectations.

With results like that, more growth is expected. Market research firm IndustryARC expects the global voicebot market to grow at a compound annual rate of 35 percent through 2023.

“Voicebots are not just used for daily operations like scheduling, placing orders, and reminding,” the firm said in a recent report.

In just the past few months, investors have taken a notable interest in the technologies. The following are just a few examples:

Business communications platform developer Dialpad raised $100 million in funding, which it intends to put to use building out its growing number of AI-based features, including voice transcription and analytics for meetings and contact center interactions with customers.

Dialpad’s new automatic note-taking feature, which uses natural language processing to flag action items and analyze sentiment, is now the foundation for Dialpad’s “Voice Intelligence” offerings. Dialpad’s technology also analyzes phone calls to identify live coaching opportunities.

Voice analytics startup Observe.AI raised $54 million in funding that it will use to develop its technologies for transcribing and analyzing call center phone conversations. Its artificial intelligence parses calls based on language and emotion, even identifying when and why periods of silence occur.