Voice Assistants in a Conversational AI Race Beyond the US
VoiceSignals #6 — Musings on Voice tech news
While the US and Europe are dominated by Alexa, Google Assistant, and Siri, there are countries that follow a different path when it comes to Voice Assistants. Tech-savvy and progressive, with a deep knowledge of local culture and language, they’re able to innovate fast with the help of large established user bases, like that of WeChat and Yandex. Among such countries is China, whose smart speaker installed base is expected to rise to 59.9mn in 2019, Japan, which is predicted to quadruple its smart speaker market size to US $38bn by 2025 and Russia, where the monthly audience of Yandex’s home-grown voice assistant Alice surpassed 30mn. If you’re wondering about the momentum let us inform you that, according to Canalys research from May 2019, China’s smart speaker shipments grew by 500% in Q1 2019 overtaking the US, achieving a 51% market share. So how did that happen? What are they doing better than giants like Amazon, Google or Apple?
China reached its large speaker shipments using festivities like the Chinese New Year’s Eve, while its dominant tech giants, the great BAT (Baidu, Alibaba, and Tencent), have used every opportunity to promote their offerings. Baidu is the country’s largest provider, Alibaba is the biggest e-commerce platform, and Tencent created WeChat, so it has access to personal data of over 1 billion users. BAT’s capabilities are comparable to those of FAMGA (Facebook, Amazon, Microsoft, Google, Apple), and obviously all tech giants are in the race for the most popular voice assistant. Add to the mix Xiaomi, one of the largest manufacturers of smart speakers, and China’s interest to dominate the AI market we should be expecting some elbow shoving. While the trade war with the US is still flaring, China is expanding its trade with the rest of the Asian market, offering voice assistants in different price ranges. But while Chinese companies benefit, culturally and geographically, from their proximity to the Asian market they do lack the environment that will attract experienced engineers and scientists, that will help them expand their technology internationally.
Russian Yandex’s voice assistant, Alice, is now being used by 45 million people monthly. Most of her audience comes from smartphone users, despite them having native digital assistants installed, while Huawei is considering adding Alice to its smartphones. According to one writer, it seems the “chit-chat feature paired with Alice’s distinctive character and quirky sense of humour is the key to the voice assistant’s growing popularity”. Soon after its release, Alice went viral on Russian social media with screenshots of funny dialogues and videos of people talking to it. Alice memes and people trying to uncover its hidden skills encouraged people to use smart assistants and disrupted the way people communicated with them. But Alice is not the only famous VA in Russia. The Mail.ru Group, owner of Russia’s two most popular social networks Odnoklassniki and VKontakte, announced it was beta-testing its very own voice assistant Marusya; while the Tinkoff Bank, a financial institution, has invested heavily in AI and speech technology for its own voice assistant Oleg. Due to the institution’s investment in high-end technologies, like voice as a biometric authenticator, it can use Oleg to identify a user’s voice, offering individualized services, without the typical hassle of going to the bank.
So, wrapping it up, according to predictions there will be over 8 billion voice assistants in use, by 2023, around the world. All the major tech players are positioning themselves to grab the most users, and obviously the most data and influence. The race has just begun and it will definitely be very interesting to watch.
What we read online…
Dear App Developer, Stop Ignoring My Feelings
Emotional Design emerged as the effort to promote positive emotions (Norman, 2007) by means of design properties of products and services. According to Van Gorp and Adams (2012), design based on emotions can affect overall user experience deeply, since emotions influence decision making, affect attention, memory, and generate meaning. The more engaged a user becomes in the game and the experience of completing the game’s tasks, the more effective the game. The same concepts have been applied to a range of applications in recent years, as developers harness the power of emotions through interactive storytelling, gamification of routine tasks, and presentation of familiar concepts in creative, engaging ways.
One of the best examples of gamification in what would otherwise be a “boring” experience is Duolingo. By gamifying the learning process with achievement points, sharing with friends, making the process more playful and colorful, the result is a 55% next day retention rate for Duolingo, on par with the average mid-grade video game, and a valuation of over $700 million. The Duolingo app also uses voice and speech recognition to help users with pronunciation. Voice has the potential to draw user emotions and utilize them within the app learning curve. Imagine when Duolingo will be able to understand learner frustration and react in a constructive way to help achieve the student’s goals. Read more >
As AI Gets More Emotionally Intelligent, So Must We
…or how to make emotion AI software work, for both sides. We need to understand AI if we want to be understood. We need to learn how it works and even ‘empathize’ with having AI in our lives. The more it enters our lives, the more it becomes a companion in the true sense of the word. “Do unto your AI as you would have it do unto you”. If you are polite and courteous with your voice assistant, the subtle inflections of your voice will register in its emotionality mainframe. This opens up your dialogue to more multifaceted interactions. Research indicates that humans want their voice assistants to be confident, yet subordinate; polite, yet productive. As with any communiqué, this is a delicate balance, requiring effort from both parties to yield the most beneficial results. Read more>
New Google Assistant & Recorder on Pixel 4
The past week we had some impressive announcements on the new Pixel 4 regarding its speech recognition technology. Google managed to reduce 100GB of speech recognition and language understanding neural network models down to less than 0.5GB. The new Assistant processes speech “at nearly zero latency,” with “transcription that happens in real-time,” and works “even when you have no network connection.” Overall, Google says the next-gen Assistant can process your query and deliver an answer up to 10 times faster than before, and with Continued Conversation support, you can keep firing off queries without needing to manually trigger the Assistant again. If a fact, it’s impressive!
What we’d love to get our hands on is the Google Recorder. Not only can speech recognition achieve up to 98% accuracy, while transcribing in real-time, the transcript is searchable even offline! This is magical and although many on Twitter argued that Apple’s iOS has similar technology, since several years, it’s a matter of focus. Apple has let “Siri shrink by clear neglect, allowing other assistants to thrive” as Brian Roemmele, from Voicefirst.expert, has put it. We’re looking forward to Google, opening up this technology to developers; it’s truly going to promise some exciting apps for our phones.
Giving 1 Million Customers an Echo Auto
Last month Nationwide Insurance announced it will give away one million Amazon Echo Auto devices to qualifying new and existing auto insurance customers. The move coincides with the introduction of new features within their Alexa Skill. The company seeks to retain customers and possibly lure in new ones while offering a ‘digital experience’ to its insured clients.
Although we’re strong advocates of companies building a digital strategy, that includes Voice, we would have liked a solution where the data would actually belong to the company and not Alexa. That would definitely give me the more value for their buck. Read more on Voicebot >
Written by Vicki Kolovou for Behavioral Signals
#voiceassistants #voice
