Speech is here to stay – Thoughts from the Conversational Interaction Conference

Feb 02, 2017

As a self-professed speech geek, I love speech technology trade shows and conferences, and over the years I’ve watched them follow a similar evolution as that of the contact center industry.  The contact center industry started with ‘Telecom’ shows, and then voice, call center, contact center, and finally many variations on customer engagement.  The speech technology industry conferences, deeply entwined with the contact center, also have morphed with the tides of where speech was being deployed. Organized jointly by Bill Meisel’s TMA Associates LUI News (Language User Interface) and AVIOS (Audio Visual Input Output Society), a non-profit speech technology professional society, the Conversational Interaction Conference, CIcon2017, is no different. Most recently this show was named Mobile Voice, which truly denoted the amount of attention being placed at the time on using voice user interfaces (VUIs) within mobile devices. This new name, however, reflects the sea change the industry has witnessed in the now broad applicability of speech and language technologies across devices, platforms, and applications, from mobile phones to virtual assistants and Bots.

Speech is here to stay.  OK. I’ve said it. And so have dozens of others over the years, and I think this time it’s fair to say we mean it.  Speech recognition and text to speech are now commonplace and accepted by a broad swath of the population.  In particular, the Millennials and later generations think of voice as commonplace. They grew up with interactive toys, games that deploy VUIs, and of course, the ever present VUIs on mobile devices.

So it is not surprising that this show was bigger and better than last year. As Bill Meisel put it, “Natural Language processing and Artificial Intelligence (AI) have matured rapidly in the past decade and have found a solid toehold”. “Connecting to computer intelligence through language allows complex interchange of information which is hard to accomplish by other means.” I heartily agree, and the linguistic and speech science community is breathing a sigh of relief.

What did this conference bring us? It brought a healthy mix of developers, thought leaders, and solution providers to discuss trends and applications using advanced technologies.  Not just speech and text, but machine learning (ML), AI, Big Data, and enhanced analytics capabilities.  The conference also highlighted design and deployment methodologies and best practices.

One of the hottest topics was the development and use of bots. Speakers from companies that are deep into the bots and virtual assistants included Google, Microsoft, Aspect Software, Interactions, Amazon, Kiwi, Inc., Nuance, noHold, Inc., and Kasisto, among others. Some presentations, such “Expectations vs. Reality: Observations of Edward in the Wild, presented by Dr. Lisa Michaud, NLP Architect at Aspect Software, focused on bots for customer service. Edward is a concierge-style bot launched at Radisson Blu Edwardian Hotels in London.  Edward was launched at a dozen Edwardian hotels with amazing success, acting as an Interactive Text Response (ITR) mobile SMS concierge service that interacts with hotel guests and handles service requests, offloading hotel staff.

However, other presenters focused on the simple joy that a bot can bring. For instance, Omar Siddiqui, CEO at Kiwi, Inc., gave some great examples of consumer engagement with Bots in his talk, “The Best Way to Bring Bots to Life”. Kiwi, Inc. was one of the launch platforms for Facebook Messenger and now delivers 50-100M messages a month to end users, focused on those moments of fun. As he said, “You can at least deliver charm, wit and an emotional moment.” Examples of the types of bots being deployed on Kiwi’s Sequel Bot Platform for Media and Entertainment include games, trivia, polls, celebrity quizzes, and many others. A few of the interesting insights he provided were that developers have found that it is typical for users to intensely engage with the bot at first, generating 200-300 messages a day, then it drops off, much as what happens when you have dialogs with new friends in person. Similarly, in a chatbot session, there might be short bursts of activity that elicit longer narratives, much like when a person gets excited or starts to rant about something. So, the bot engagement behavior with users is not always a simple back and forth response, that other customer interaction channels have shown. Omar made a strong point that it’s imperative with the differences in user behavior than typical customer interaction that developers need to heed best practices in design. Examples included:

  • Focus on a particular use case/job that can serve directly and well
  • Avoid long stretches of talking at users; generally insert interaction every 3-4 minutes and be ready for interrupts. Be careful.
  • Remember your audience and social communications context of the bot.

One of the most enjoyable presentations was “The Future of Conversational UI” presented by Vera Tzoneva, Global Product Partnerships, Google Assistant. Vera spoke of some of the partnerships that have fed into the development of some pretty cool virtual assistant apps, from partners such as CNN and Buzzfeed.  These have resulted in everything from Google Assistant accessing Genius, a partner that can identify song lyrics for users, to Headspace, which provides users with access to guided meditation. In the instance of the later the user simply says “Google, I want to meditate”, and it will come back and ask the person if they want to talk with Headspace, and if they say yes, it connects them and gets them to the desired meditation.

Overall, whether talking about the efficacy of deploying virtual assistants to offload customer service agents, or launching VAs and Bots to better engage and even entertain customers, the conference presenters highlighted how far technically we have come since the early days of speech-enabled IVR.  And the good news is that we are only just starting with much more to come.


Nancy Jamison


Nancy Jamison is a Principal Analyst within Customer Contact within the Digital Transformation group at Frost & Sullivan. She covers all aspects of customer contact including cloud and premise-based systems and applications in the core areas of inbound/outbound routing, IVR, Workforce Optimization, and recording and analytics, with a particular focus on peripheral and emerging areas that impact the Customer Experience. These include speech technologies, omni-channel customer care, Big Data, digital marketing, Back Office Workforce Optimization, and Support Interaction Optimization.

Add Pingback
blog comments powered by Disqus