Exploring Voice Assistant Technology in Africa
Technology is disrupting economies globally, more so in African countries. And positively so. Having lagged in innovation in the past, African countries have been better placed this time to leverage technology, allowing them to leapfrog several development stages that are typically de rigueur in other spheres.
How should companies doing business in Africa or looking to venture into the continent take advantage of technology? They should embrace it certainly. Traditional means of advertising are not likely to be as effective as they once were. With increasing internet access, African consumers can also be reached directly for product ideas, testing, and feedback.
The telecom, retail, banking and probably every industry have been in some way affected by these disruptive inventions which have drastically changed the places people consume content and do business. One such unexpected but disruptive trend in recent times has been the slow but steady rise in the use of Voice Assistants.
Tech Target defines voice assistants like this:
“A voice assistant is a digital assistant that uses voice recognition, natural language processing, and speech synthesis to provide aid to users through phones and voice recognition applications.”
Voice assistants are digital applications operated on smartphones and other devices that understand human language and try to operate for the user in a way similar to how a human companion would do.
These technologies, powered by Artificial Intelligence (AI) aim to make information more readily available to consumers in order to simplify and expedite their daily tasks. As voice-based interfaces become ubiquitous, more people are using smart speakers to shop, set reminders, and get answers to simple but essential questions like the weather.
In the consumer space, the dominant voice-based virtual assistants (VAs) are Google Assistant, Apple’s Siri, Microsoft’s Cortana and Amazon’s Alexa. Siri was the first — making her debut on the iPhone 4S in 2011. But it was Alexa that arguably kicked voice uptake into high gear.
Research shows half of all searches will be voice-based by 2020, and this massive pivot towards voice commands is set to create an entire ecosystem of applications and interactions.
After all, the Virtual Digital Assistant (VDA) worldwide market size is currently estimated at $5.21bn, a number set to be slightly under 2 billion people when projecting the number of digital assistant users worldwide by 2021.
The African Narrative
Already, today, quite a number of smart speaker applications are being either imagined or developed for various markets. This is especially important in a market like Africa’s, where its digitally competent consumer base is distributed across the different income levels.
Analysts have been hailing the rise of voice or speech as the next major interface trend for a few years, but what are the implications of a shift from screen to voice in the African consumer space?
Here’s a rundown of some of the implications that relate to this development highlighting the challenges and the prospects that could be implemented to offset them.
The Accent Problem
Challenges: These technologies are built with the average US consumer in mind, so they often misinterpret communications made to them in other accents.
According to a 2018 study in the UK, 79% of people with regional accents regularly change the way they speak so that digital assistants like Alexa or Siri can understand them.
Slowly and deliberately pronouncing words with an American or British accent while interacting with voice technology is something Nigerians with Nigerian accents can relate to. That is because technologies like Alexa, Siri and Google assistant understand English, but tend to choke on Nigerian, Kenyan or other accents.
The Washington Post reported that Google and Amazon admit “accents remain one of their key challenges — both in keeping today’s users happy and allowing them to expand their reach around the globe”.
Prospects: Though the accent problem is still a huge downside to voice technology, corporations all over the world are beginning to adapt their products and services for the African consumer.
Kola Tubosun worked with Google between 2015 and 2016 on a Nigerian English voice project. In November 2018, the Google language team talked to developers and users in Nigeria about their renewed interest to build capacity for local language interaction with Google systems.
According to him, “when the project is launched on Google’s various services in the nearest future, it will be of great help to Nigerian users who don’t speak like Americans nor need to do so.”
A recipient of the 2016 Premio Ostana “Special Prize” for his work in language advocacy, Kola says the need to modulate our accents to help digital assistants understand us “is replacing the social pressure we already feel when we go abroad, to change our speaking, to conform to other people’s expectations.”
Kola created Yoruba Name, an open-source project to preserve and document all Yoruba names in a multimedia format. Yoruba Name released the first online Text-to-Speech for Yoruba in 2017.
The Language Divide
Challenges: African languages also lack enough data to train machine learning systems, disincentivizing researchers to start from scratch.
The established voice recognition technologies mostly work on listening to what the user is saying and then putting that into text format (text to speech). Differing manners for which this can be done exist, however, this is the prevailing way that voice recognition works.
Recent technologies take this one step further and offer voice assistants which not only figure out what you are saying but then they research massive databases of information to help get answers to questions that you might ask. You see this integrated into smartphones (Siri, Google) or in-home speakers (Echo with Alexa).
Research has it that building accurate machine translation services requires approximately 100,000 hours of recorded speech.
Even though languages like Swahili are spoken by almost 100 million people, there isn’t a well-known repository that can be used to feed speech recognition software.
The multilingual nature of modern Africans also means developers aren’t processing mixed language models. Code-switching is inherent in many countries, with users either speaking or writing in two or three languages in one post or conversation.
An example of this is the combination of English and Swahili in Kenya’s Sheng, the mixing of French with Arabic or Berber in Algeria, or even the Romanization of Amharic.
Prospects: Companies are now taking steps to train systems on new languages and accents. In 2018, Amazon released Cleo, an Alexa skill that allows users to teach the AI assistant new languages.
Artificial Intelligence-based technologies are emerging which are eliminating the need to convert voice to text and then refer to a massive database when dealing with understanding the intent of the user who speaks the language.
If one can eliminate the need to refer to massive databases to examine every word to determine intent, it can allow for artificial intelligence to transform voice recognition. This can then allow integration of voice recognition into not only home assistants but into other everyday products that are not connected to the internet that are in use in any language, quite easily.
Even the ability to call your mobile phone operator and get top-up airtime, find out where to change a SIM card, pay your electricity bill, or other services which are not currently available in all languages, could be suddenly available.
Artificial Intelligence has the power to use logic and compute massive amounts of data or develop strong algorithms to make this a reality and it is not as far away as we might think. While AI has the ability to affect much more than voice recognition, this one advance might end up having a major effect in helping all people be understood.
Challenges: The main problem is the way that current technologies work and how this can be scaled globally. While American English is a considerable market, in the hundreds of millions of people, it represents less than 5% of the global population so it is not really solving the global communication problem.
It is said that a major manufacturer took about two years to launch its smart home product in the German language. This is because it takes thousands of hours of speech and hundreds of speakers to be able to have a database large enough to convert text to speech and then derive intent from it.
This is for a massively used language like German which has a consumer market ripe for products like personal assistants, but what about languages like Hausa or Swahili where the user subset is not as large or as potentially lucrative?
Research by the early-stage accelerator Digital Financial Services Lab and research consultancy Caribou Digital shows developers are focusing much of their efforts on improving English language capabilities and less on languages from developing nations including Africa.
Even though companies have made enormous advances in Natural Language Processing (NLP), or the ability of computers to comprehend the human voice and language, there’s an emerging divide that relegates low-income populations and less widely-spoken languages to the background.
As a result, this disparity could not only create a gap between those who can use this form of artificial intelligence for communications and those who can’t but also hinder users’ abilities to exploit these applications for development interventions like healthcare and finance.
Prospects: Africans are increasingly connected, with high mobile penetration — reaching one billion in 2017. By 2030, Africa will have 16% of the world’s internet users, which reflects the growth of over 260% from 2017. This makes Africa the fastest-growing region for the number of internet users, creating compelling opportunities, such as in mobile banking and online retailing.
With the “Africa Rising” narrative, the emphasis has been on a rising middle class. However, untapped potential still exists at the “bottom of the pyramid” (BOP). The BOP consists of households with an annual disposable income of USD2,500 or less.
South Africa is unique compared to the rest of the continent, as the country’s income divide is the largest in the world. In terms of spending, its BOP discretionary expenditure is the highest for the region and is set to increase to 46% in 2030. Nigeria is one of the most important BOP markets, thanks to its large size in both proportional and absolute terms.
BOP in both Nigeria and South Africa is not expected to shrink through to 2030; thus, companies should adopt a stable, long-term BOP strategy for these countries.
Africa will be home to some of the world’s fastest-growing economies in the coming years. Ethiopia, Rwanda, and Mozambique are set to record the fastest growth in real GDP terms, with growth rates that are well above the global average.
Africa’s two largest economies, Nigeria and South Africa account for nearly 50% of the continent’s GDP in 2017. However, by 2030 these two countries will represent just 37% of Africa’s total GDP, demonstrating the rising economic importance of Africa’s emerging markets.
Businesses that can win the loyalty and trust of today’s Kenyan BOP consumer will be rewarded with an emerging middle-income consumer base with higher purchasing power in the future.
Just like MTN did in the Nigerian telecommunications market, the retailers who run ahead of the trend to start developing skills in this area, creating ads and content to capture the tiny audience the technology already has in the country and most importantly, bring it to the eyeballs of the consumers like Amazon did with its Alexa will most likely have the lion’s share.
Challenges: Research by Maintel recently unveiled that many businesses see the possibilities offered by virtual assistants, but they still have many obstacles to overcome as far as consumer opinions are concerned; the biggest issue being, trust.
Consumers are either uncomfortable with the amount of data that service providers collect on them, or don’t trust them to keep their data safe or use it properly. Data anxieties are also reflected in what customers are willing – or unwilling – to use virtual assistants for.
Almost half (48 percent) would be uncomfortable using these channels to buy a product from a retailer, and 63 percent would be afraid to use a virtual assistant to buy from an insurer or to submit an insurance claim. Another two-thirds (66 percent) would hesitate to use one to transfer money between bank accounts.
These stats will be a major concern for any brand looking to unveil services using these virtual assistants, particularly those operating in sectors that handle more sensitive data, such as retail, finance, or healthcare.
Prospects: Despite this consumer concern, it’s reassuring to know that businesses appear to recognize the issues they face. More than a third (36 percent) think that the biggest barrier to consumer adoption is security fears, a quarter think consumer trust is the largest hurdle, and 22 percent believe that privacy concerns are the most important factor.
Essentially, they need to build trust in the capability and security of virtual assistants. So, what are some of the practical steps companies can take to build this trust?
Currently, this technology is potentially open to hackers due to vulnerabilities such as biometric spoofing, where attackers use physical identifiers, including voice or fingerprints, to impersonate their victims. As the technology matures, it’s likely these potential breaches can be prevented. However, it will take time for developers to iron out these issues.
In the short-to-medium-term, companies must take action to assure customers that security is very much a priority. Businesses may consider implementing an extra layer of verification to the process. For example, if a customer begins an interaction using a virtual assistant embedded in a mobile app, they could be sent a text message with a unique code to confirm their identity.
Of course, one of the major benefits of virtual channels is their speed of resolution – customers interact with the services in real-time, without needing to wait in a queue to speak with another human – so businesses need to carefully balance the number of authentication methods they deploy to maintain the greater speed and convenience of the customer experience.
Since the Cambridge Analytica scandal, many consumers are also highly dubious of companies collecting data. To assuage these fears, businesses need to make it abundantly clear to consumers what data they will be collecting via virtual assistants, how this data will be kept secure, and, perhaps most importantly, what this data will be used for.
This can certainly be relayed to the consumer via a message through a virtual assistant, but it should also be displayed via FAQs on the company website and details sent directly to the consumer by email or text.
Challenges: According to digital research from Deloitte, Virtual Assistant use is still relatively low in South Africa and the penetration of connected home and internet of things devices is low and “in the early adoption stage”. According to its “Global Mobile Consumer Survey 2017: The South African Cut” report, only 5% of SA survey respondents reported owning voice-assisted speakers, such as an Amazon Echo.
The majority of Nigerians who actively use the technology now are those who live in the diaspora and recently moved back home or are just here visiting. Mainly because it has somewhat become a part of their daily lives due to its wide usage in the environment from which they may have returned.
The remainders of its substantial amount of users are perhaps just still intrigued by the technology and thus often use it for fun.
The use of voice assistants still seems deeply few and far in between just as mobile phones and laptops were left as exclusive preserves for the luxurious some years ago.
Prospects: According to its “Global Mobile Consumer Survey 2017: The South African Cut” respondents were asked about the use of Virtual Assistants on their smartphones, and results are dis-aggregated by age.
So, for example, those in the 16-24 age group report using their VAs for playing music (25%) and for setting alarms and reminders and creating calendar entries (16%) rather than for navigation (13%). This indicates the comfort in using these tools in everyday life. Users aged 35-45, on the other hand, report navigation as their primary use (23%).
Nigerians who make transactions on e-commerce websites like Jumia and Konga now conveniently order goods and services with their gizmos in a manner bearing a huge similitude to and with the ease of a corporate executive requesting coffee from his secretary.
However, just like the earlier trends in the use of mobile phones and laptops did not continue, the current rate of usage of voice technology will soon become history.
This means ordinary users are going to grow accustomed to interacting with online material through their voices. If you want your brand to stay relevant in the long term, you’ll need to adapt.
In some cases, that means optimizing your content to better address simple user questions (so you can be featured in the rich answers at the top of Google’s search rankings). In others, that means simplifying your content so it can be easier to engage with through smart speakers or mobile devices.
Most companies that build smart speakers still haven’t made them readily available here in Africa, with the general perception of the continent as a potential market still largely prevalent. However, with the new problem-solving solutions they can bring to Africa in general, smart speakers can be the next flag bearers of a technological revolution.
While it may not seem like the urgent need of the hour in Africa, understanding the capabilities of these devices make them appealing. With home automation and IoT poised to grow in the future, a piece of technology that makes our lives easier is not a hard sell.
With more features including “visual responses” and a willingness to adapt their products and services for the non-American user, voice assistant technology has the potential of being a revolutionary phenomenon that changes the perception of IoT in Africa.
Just like the earlier trends of mobile phones, laptops — and most recently wearable tech — the current rate of usage won’t remain low forever. In the not too distant future, Africans will be talking to home assistants that not only help with task-oriented commands but also goal-oriented ones.
- Alexa, are you a salesperson? How voice assistants could influence purchase decisions. (2019, October 25). Retrieved January 4, 2020, from https://econsultancy.com/how-can-voice-assistant-alexa-influence-purchase-decision-for-fmcg-products/
- Is Africa Ready for a Smart Speaker Revolution? (2018, August 10). Retrieved January 4, 2020, from https://medium.com/@SoftcomNG/is-africa-ready-for-a-smart-speaker-revolution-8be11f9311e5
- The Impact of AI in Voice Recognition in Africa. (2019, March 17). Retrieved January, 4 2020, from https://medium.com/datadriveninvestor/the-impact-of-ai-in-voice-recognition-in-africa-53e5ff58f5d9
- African languages are being left behind when it comes to voice recognition innovation. (2018, November 27). Retrieved January 4, 2020, from https://qz.com/africa/1475763/african-languages-are-lagging-behind-when-it-comes-to-voice-recognition-innovations/
- Make way for the smart speaker revolution. (2019, February 14). Retrieved January, 4 2020, from https://www.businesslive.co.za/fm/fm-fox/digital/2019-02-14-make-way-for-the-smart-speaker-revolution/
- Digital trends that can help change the game for small South African businesses. (2019, April 16). Retrieved January, 4 2020, from https://www.bizcommunity.com/Article/196/16/189851.html
- Euromonitor International (2018). Shifting Market Frontiers Megatrend Africa Rising. Retrieved from https://www.euromonitor.com/shifting-market-frontiers-africa-rising/report
- Voice assistants, the future of online retail in Nigeria (1). (2018, February 5). Retrieved January 4, 2020, from https://wildreams.co/voice-assistants-the-future-of-online-retail-in-nigeria-1/
- If We All End Up Sounding Like Americans, You Can Probably Blame Voice Assistants. (2019, January 24). Retrieved January 4, 2020, from https://techcabal.com/2019/01/24/if-we-all-end-up-sounding-like-americans-you-can-probably-blame-voice-assistants/
- Alexa, How Can I Build Consumer Trust In Virtual Assistants? (2019, November 25). Retrieved January 4, 2020, from https://data-economy.com/alexa-how-can-i-build-consumer-trust-in-virtual-assistants/