What's next for speech recognition?

What does 2017 have in store for speech recognition?

Technology is engrained in our lives. We are always looking to innovative technologies for ways to make everyday tasks simpler and more efficient. The next 12 months will be no different.

Here are four areas to keep an eye on in 2017:

Security of the data

The debate around data sharing and concerns over cloud security continues, particularly in light of The Information Commissioner’s Office recent investigations of some key UK household brands. Fortunately, the technology protecting organisations’ data is continually evolving to be more robust and secure.

This ties into the debate around how organisations are managing the process in-house by moving their own data to private clouds, whether it is to mitigate security risks, or to save costs in the case of AstraZeneca and the DVLA.

While it is unclear whether the trend for managing technology in-house will remain, technology providers know they need to adapt to ensure their customer’s needs are met both in the public and private cloud.

Managing the skills shortage

Machine learning (ML) requires a certain skill set, one for which there is a shortage of experts and engineers.

One of the biggest challenges is that universities cannot provide the training and education in this area at the rate it is needed. However, using research papers and online courses available to them, tech companies are investing time and resources into training their staff to understand ML and how it can be applied to their business, from analysing and understanding customer conversations, to wearable devices for speech disorders. Those that continue to invest in this training over the next few years will reap the benefits of well-trained staff which can innovate speech technology.

ML and human understanding

With more articles and research being published around ML, so the awareness, investment and interest in ML applications will continue to grow.  The combined wealth of knowledge means we are continually building more powerful systems, typically by innovating on neural network architectures, and applying more advanced models of the world into everyday dialogue tools such as virtual personal assistants and speech recognition software.

Keeping up with demand on mobile

Mobiles have become an integral part of daily life. With two thirds of people now with smartphones these devices have become the first port of call, for many, for everyday activities over desktops and laptops.

Mobile creators will continue to turn to Graphics Processing Units (GPUs) to improve device capabilities. GPUs are getting faster and stronger and can use more algorithms which are enabling more powerful mobile applications. Many apps are now able to really harness neural networks, which are very well suited to GPUs, and allow us to use more tools, like speech recognition and image processing, to manage the everyday jobs we do.

Tony Robinson, CTO of Speechmatics

Speechmatics will be located at Hall 7 Stand 7C72 at the Mobile World Congress.