#2 – Speech Recognition goes mainstream, but Europe is left in the dark

In 2017, Automatic Speech Recognition (ASR) will approach maturity and go from hype to everyday life, from a niche market technology to mainstream adoption. It will have broad applications and a huge impact on both smartphone and IOT devices. However, in stark contrast to historical trends, Europe will trail behind the US and China in the adoption of this innovation. This will be due in part to language barriers, but also to the lack of investments and the slow pace of innovation within the European local markets towards the growth of ASR technology.


In 2016, ASR hit a symbolic milestone with a system accuracy on par with professional human transcriber levels (e.g.,Microsoft ASR reached 94.1% accuracy rate). In the last couple of years, the global ASR market has grown tremendously. Major technology giants such as Google, Apple, Baidu, IBM, Microsoft, Amazon and Intel have all established a strong presence and ecosystems in the ASR field through software (Microsoft Cortana, Baidu Yuyin, IBM Watson), combining software and hardware (Amazon Echo, Google Home, Apple Siri) and ASR APIs (Google Cloud Natural Language API, Apple SiriKit). Northstream believes this trend will continue with increased investment, broader applications and an expanding ecosystem. More mobile operators may also join this sphere, similar to what China Mobile has done with its ASR API platform Lingxi, which they have deployed in their own customer call centers in China. Also, the penetration of 4G and the development of big data and cloud computing provide a solid technical foundation for ASR’s commercialization and popularity. These developments herald the arrival of the ASR application era.

The rapid innovation and commercialization of IOT is supporting the growth of ASR, especially in the domain of smart homes (Amazon Echo, Google Home, Ding Dong), wearables (Apple Watch, Pilot earpiece translator, Sony Xperia Ear) and smart cars (Roewe RX5 Internet car, Carrobot). In 2015, 95% of Apple watch users reportedly used Siri, of whom 50% used it daily and 31% used it several times a day or more. These numbers greatly exceed the frequency of Siri usage among iPhone users. ASR complements IOT devices particularly well due to its low implementation cost, its quick response time, the fact that it doesn’t require a screen and because a convenient, touch-based interface is difficult to manufacture on such small devices. As a result, Northstream predicts that ASR will highly influence and empower the design and use cases of IOT devices in the future.

On the consumer side, ASR usage has increased drastically due to the increased satisfaction levels, ease of use, and reduced task time. This is especially the case for consumers in the US and China. Google claims that 20% of its searches are through ASR. However, ASR usage and satisfaction levels remain relatively flat in Europe as progress in languages other than English and Chinese has been limited. The lack of significant ASR investments for languages other than English and Chinese by major technology giants and the absence of major local players may explain the slower development and penetration rates in Europe. In the past, the Nordics have consistently been early adopters of new mobile technology. However, due to the language barrier, they are likely to lag behind when it comes to adopting ASR. And by extension, so will Europe.

ASR is improving quickly, but to realize its greater potential, it needs to be combined with other AI technologies, such as natural language processing and deep machine learning. Even though singularity may still be some way off, we can expect a broad application of ASR and that it will have a great impact on the design and development of future smartphone and IoT devices.

Next read

#3 - The NFV hype is over, now it will really take off

Northstream predicts a breakthrough and strong growth in NFV activities in 2017. The incentives for operator implementation are high as some new services (e.g. VoLTE) will benefit from virtualized network functions and some future ones (e.g. network slicing) are dependent upon them. As a result, NFV in 2017 will be driven by services such as [...]
Read more