Everything You Missed From Our Voice Technology Event

Five great speakers:

• Andrea Muttoni, one of Amazon Alexa’s leading Tech Evangelists

• David Attwater, Senior Scientist at EIG

• Liz Barber Client Director at RADA Business

• David Low, Executive Product Manager

• Lisa Vigar, Senior Product Manager from the BBC Voice and AI teams

A real mix! On the night, each speaker brought new ideas, examples and insights from quite different perspectives – including an audience ice breaker activity that made this year’s voice technology event considerably louder.

Andrea Muttoni, Amazon

“Alexa is now everywhere. These days it’s not just dedicated devices built by Amazon and others – all sorts of manufacturers (from cars to kitchen appliances) are integrating voice so that Alexa, and other voice assistants, really are everywhere.”

• Voice commerce: The recent development of voice commerce has enabled us to design for full skill interactions. It’s a new way to design for voice – it’s flexible, leading to better implementation.

• A new UX challenge: Amazon have created two ways of monetising skill experiences: Amazon Pay and In-Skill Purchasing. Whether it be a charity donation or train booking, Amazon Pay can manage it smoothly and easily, whilst In-Skill Purchasing is for more premium content. It’s allowed voice to venture into the creation of virtual gaming products, such as add-on content and the unlocking of features to continue and enhance the immersive experience.

• Situational design: To improve the way we design for voice, Amazon is moving away from nested menus and flowcharts, over to situational designs. These enable full skill interactions to be displayed – like an in-depth story board.

David Attwater, EIG

“Any designers considering social and emotional relationships must be very careful. The machines need to meet with consumer expectations, and – although we’re good at language – we need to realise just how complicated it is to make it feel natural.”

• Emotional response: Designers must consider the social and emotional responses we can create using language. Think body language, context, relationships, social connection and verbal pointing – then imagine a conversation without these aspects. Language is incredibly powerful, able to take symbols that mean one thing, and put them together to create a new meaning.

• Uncanny Valley: If designers get too close to nailing the characteristics of a human in a machine, and then get it wrong, it can actually initiate a disgust response from that person, who will then completely disregard the machine. Machines are intelligent, but humans even more so.

• Creating naturalistic dialogue: We know when to speak and when to be quiet, how to retain information and not deluge people – but when it comes to translating this into voice tech, the tools are really limited. If we’re going to fake it, the execution is critical.

• Adapt and survive: Use the intelligence of the user and get them to adapt their behaviour to suit the machine, rather than to try to program the machine adapt to suit the user.

Liz Barber, RADA Business

“As humans we consciously and unconsciously adapt our voice and change it to suit the situation. Should we be trying to replicate this process of vocal production unnaturally?”

• The conscious unconscious: We unconsciously and consciously alter the way we use our bodies to create and engage with sound. Liz kicked off our voice technology event, asking the audience to pair up and simply count “one”, “two”, “three” alternately between them, and repeat. Only it’s not that easy. She then made the task progressively more difficult, introducing claps, clicks and gestures in place of numbers. Returning to the original task was suddenly a lot easier. Why? Because as soon as there is an action to perform, it distracts from the pressure we feel to create sound. In fact, adapting our voices is very natural and helps us to evolve in our vocal production.

• Voice evolution: Technology has begun to impact this vocal production and natural voice evolution. Replicating the process we go through to construct a conversation with a human, with a machine, is no easy task when there are a huge number of physical factors to consider – how can you design for a psychosomatic interaction? Creating a dynamic relationship between mind, body and environment, and then slipping a voice assistant into the mix, is a challenge facing all voice designers.

David Low and Lisa Vigar, BBC

“After launching, we soon realised that we needed to design for real-world users”

• Complications of voice: Making day-to-day tasks easier and quicker via voice is a lot easier said than done. Search terms had to be classified to ensure they suited the way a user would actually interact with Alexa, e.g. people wouldn’t ask for ‘Radio One’s Dance Party with Annie Mac’, they’d instead ask for ‘Annie Mac’ or ‘Dance Party’.

• Staying agile: Despite seven rounds of user testing of a children’s skill, and launching confidently, users didn’t come back to use it. In fact, only 70% of people were using it once a week. The BBC team acted fast; checking reviews, asking questions, setting goals and challenging statements. This helped to quickly identify areas of the skill that needed enhancement and drove repeat usage.

• Really know your audience: Designing for kids isn’t like designing for adults. Their minds are way more impulsive and they tend to answer questions posed to them immediately, rather than waiting to hear all the options before making a decision.  Questions were switched around to give the options first, answer last – so they had to listen to the full exert. This principle applies to anyone designing a skill – you might think you know the behaviours of your audience, but you can’t assume that you understand how they translate for voice.

At last year’s voice technology event, speakers focussed on the reach of the tech and the devices. It’s clear that voice is here to stay as this was barely mentioned. What’s clear is that now the challenge is for agencies, organisations, developers and designers is to create skills that understand both the limitations and opportunities of machines.

To paraphrase David Attwater – remember, voice assistants are a tool, not a friend.