pogonici - stock.adobe.com
Microsoft’s purchase of Nuance for $19bn represents a major bet for Microsoft as it targets the emerging voice technology market on multiple fronts.
First, it represents an investment in healthcare, a sector that IBM has been pioneering with artificial intelligence (AI) for many years. Google has also been steadily expanding its healthcare footprint, such as through the Nightingale partnership with Ascension, one of the largest private healthcare operators in the US.
As lockdown measures begin lifting, the IT industry sees a huge opportunity to push forward the merits of digitisation to enable businesses to grow. Transcribing human speech from customer conversations that occur in a call centre, or between a clinician and patient, or a financial advisor and a client, represents an untapped opportunity for digital transformation in many businesses.
During an investor call covering the acquisition, Microsoft CEO Satya Nadella said: “First and foremost, this coming together is about empowering healthcare. It’s about applying the technology, talent, industry expertise and partner mindset of our two companies to this very critical sector. Together with our partners we will put advanced AI solutions into the hands of professionals to drive better decision-making and create a more meaningful human connection. AI is technology’s most important priority and healthcare is its most important application.”
Microsoft positions Nuance as providing AI at the edge – in this case at the point of delivery when a patient receives a consultation from a clinician. Nuance offers consumer, business and industry-specific speech-to-text software to convert human spoken words into a machine-readable format. It also has products in the intelligent assistant space. In the consumer market, products such as Amazon Alexa, Google Home and Apple’s Siri have enabled people to speak to devices and have their queries relayed back via text-to-speech. While Microsoft’s own offering, Cortana, is built into Windows 10, industry watchers regard it as less successful than the more established rivals.
“First and foremost, this coming together is about empowering healthcare. It’s about applying the technology, talent, industry expertise and partner mindset of our two companies to this very critical sector”
Satya Nadella, Microsoft
Lian Jye Su, artificial intelligence and machine learning principal analyst at tech market advisory firm ABI Research, said: “Microsoft has attempted to develop its own conversational AI in the past through Cortana. The company also acquired conversational AI startup Semantic Machines in 2018. However, all these have yet to allow Microsoft to replicate the success of Amazon, Google and Apple in the consumer space, and IBM in the enterprise space, and to compete directly with them.”
According to Jye Su, the Nuance acquisition demonstrates the value of conversational AI in its future roadmap.
“Intelligent assistants have moved into a new phase of adoption as thousands of firms employ chatbots, voicebots or virtual agents to improve customer experience and employee productivity,” said Dan Miller, lead analyst at Opus Research. “Their challenge is to make intelligent assistants proficient, scalable and omni-channel. Nuance scored highly with solutions that include a flexible, customisable platform that is open, combined with professional services that allow enterprises to employ a mix of AI and live agents.”
During the investor call, Scott Guthrie, executive vice-president of cloud and AI at Microsoft, described how Nuance could be used to power ambient clinical intelligence, integrated with Microsoft Teams. As an example, he described how a radiologist’s voice notes of a medical scan could be captured and analysed alongside image analysis of the scan.
Nicholas McQuire, chief of enterprise research at CCS Insight, said the structural change brought on by the pandemic meant enterprises across the board were demanding more from the cloud giants, particularly when it came to domain expertise and industry solutions. This has ultimately forced Microsoft’s hand in the wake of its launch of Microsoft Cloud for Healthcare.
Nicholas McQuire, CCS Insight
“In the past, we have seen firms like IBM buy industry specialism in datasets, but Nuance delivers Microsoft a more mature set of AI solutions in areas such as speech recognition, document processing, fraud detection and image recognition,” said McQuire. “Ultimately, these will prove key to differentiating Azure to healthcare customers against its largely horizontal competitors.”
Beyond healthcare, Microsoft sees other opportunities in professional services, such as where a financial advisor is able to combine information from a conversation with a client with product and market information.
By 2025, analyst Gartner forecasts that 40% of all inbound voice communications to call centres will use speech-to-text technology. In its Market guide to speech-to-text solutions, Gartner predicted the market would evolve in two directions over the next five years. First, niche products will continue to play a role in supporting common languages (for example, Malayan or Sinhalese) or applications targeting specific niche requirements.
“Meanwhile, broad suites from the very large artificial intelligence cloud providers will increasingly dominate the natural language translation ecosystems. These will share language and acoustic assets for broader performance across speech-to-text, text-to-speech, speech mining, translation, conversational platforms and natural language generation,” the report’s authors wrote.
Forrester principal analyst Kjell Carlsson said voice analysis, where the AI is tasked with understanding a human conversation, is the hardest of the AI capabilities to develop internally. Over the past year, this has led to a number of acquisitions, such as Medallia’s purchase of Voce.
“Microsoft is looking to take the lead in creating a second-generation AI application platform with cutting-edge edge computer vision (Azure Percept), text-based NLP/NLG (GPT-3 licensed from OpenAI and its internally developed Turing-NLG), and now voice (Nuance). Now it just needs to boost its automated machine learning (on tabular data) capabilities,” added Carlsson.