Sound Thinking about Detection and Recognition

by | Wednesday 29th August 2018

The growth in the voice recognition market has been one of the greatest technological achievements in recent years.
Despite some obvious hurdles – nuances of speech, regional accents, background noise, and the psychology of talking to your device – end users are reportedly very happy conducting searches, turning on lights and adjusting the heating simply by asking for it.

In hindsight, this vocal interaction with technology is a natural progression from inputting data via a keyboard.
And yet, while any vision of the future would have humans talking to their AI counterparts, why would listening only be confined to speech?

Why Monitor Sounds?

It makes perfect sense that any constantly listening device would monitor sounds other than human voices; just as humans interact with the noise our surroundings give off, why shouldn’t our phones, cars, or smart home devices do the same?

Furthermore, this ability would be of enormous benefit in a wide variety of applications.

Most significantly, the ability of devices to interact aurally with their surroundings can bypass or augment other forms of technology.

Applications for Sound Detection and Recognition

Take security for example. If your smart home device can listen to your surroundings while you sleep – picking up and processing the sounds of glass breaking, car alarms, angry voices, and unexpected movements inside and outside the home – it can determine whether or not the home is under threat.
Furthermore, if that ability is used in conjunction with existing technology – movement sensors, heat sensors and the like – it makes it more reliable, meaning that you are less likely to be woken by a fox crossing your drive.

Sound recognition can also be effective where the hardwired alternative is problematic. Take factories for example. Machinery manufacturers and suppliers increasingly offer a diagnostic and maintenance service to customers, which often requires sensors and cameras linked to the supplier’s head office, so that an engineer can monitor the efficiency of those machines. While this reduces unnecessary time spent driving between locations, it also means that suppliers can predict when a key piece of machinery needs to be replaced.

Such integration forms part of Industry 4.0 – or the fourth industrial revolution.

However, mechanical sensors have their limitations, and not all wear and tear can be monitored and predicted. Which is where sound recognition can come into its own. The sounds and vibrations a machine emits can indicate an imminent failure where visual checks can’t, thus potentially saving a company $thousands in unplanned downtime.

These are just two areas, but SAR Insight & Consulting predicts growth in other markets, such as personal wellbeing apps, automotive, healthcare, insurance, and many others.

Market Growth

The technology supporting sound recognition is still in its early stages, which means there are not only opportunities for those companies established in the voice recognition arena, but for start-ups specializing in this sector.

According to our latest piece of research – Sound Detection & Recognition – this market will see significant growth over the next five years. SAR expects more than 600 million devices to ship with sound recognition in 2023, totaling over 1.3 billion devices over five years.

Contact the author:

Peter Cooney