Headphones gain AI-powered 'sense of hearing' via new Audio Analytic patent
09 April 2019
A new granted patent from Audio Analytic, a leading provider of sound recognition software, aims to bring an artificial intelligence (AI) powered 'sense of hearing' to headphones, earphones & earbuds.
The new patent enables these ‘hearable’ devices to recognise sounds that cannot be heard by wearers, as well as adapting to the acoustic environment to enhance the audio experience and improve battery life. As a result, consumers can lose themselves in their music without losing touch with the world around them.
Today, a large number of headphones, earphones and earbuds are equipped with external facing microphones in order to offer active noise cancellation. The same microphones can also be put to better use for intelligent sound recognition. Giving these devices a human-like sense of hearing empowers brands in this market to offer consumers new and compelling contextual-sensing benefits, such as alerting them to undetectable sounds while wearing hearables or allowing select sounds through – such as a siren or alarm.
In addition, by better understanding the acoustic surroundings, headphones, earphones or earbuds could also enhance sound quality and better manage battery power by automatically choosing when to turn noise-cancelling on and off or by modifying the equalisation settings in different environments.
The new patent can be viewed here: patentimages.storage.googleapis.com/21/07/6d/d556292a4f412b/US20180233125A1.pdf
The patent identifies four key opportunities this technology creates, which both brands and consumers can benefit from.
1. Hearables can understand important sounds happening in the world around a wearer, alerting them to potential danger or important audio events by increasing transparency or pausing music.
For example, a consumer wearing a pair of true wireless earbuds while running can be warned of an approaching emergency vehicle. Or somebody at home wearing over-ear headphones can be notified if somebody is knocking at the door, or the sound of a carbon monoxide alarm.
2. The patent covers the ability for headphones to understand when the wearer is in a conversation, automatically changing settings, such as pausing music or increasing transparency to enable a conversation to take place without the user having to physically interact with the headphones or app.
3. Headphones can dynamically turn noise cancellation on or off, according to the acoustic environment – rather than having to be manually turned on and running at full power all the time. This means that headphones could offer better battery life for consumers.
4. And finally, headphones or earphones can also offer dynamic EQ, adjusting equalisation settings according to the acoustic environment, enabling devices to ensure the user gets the best audio experience possible, no matter where they are.
Dr Chris Mitchell, CEO & founder of Audio Analytic, commented: “The hearables market is a very exciting place right now. Consumers are replacing wired headphones with both wireless and true wireless devices, creating the opportunity for wearers to express individuality and enjoy high quality audio experiences in a form factor that suits their lifestyle needs. The surge in demand for untethered devices is driving innovation and competition, and the major brands are looking for sources of uniqueness and differentiation. As with most other sectors of the consumer electronics market, they are tending to look towards AI – and we see sound recognition as a major part of that movement.”
As well as headphones, earphones or earbuds, Audio Analytic’s patent also covers other wearable devices, such as fitness trackers, smart watches and smart clothing.
About Audio Analytic
Audio Analytic is a global leader in intelligent sound recognition, using advanced, edge-based AI to provide consumer technology with a wide sense of hearing.
At the heart of the company is a technology platform made up of two synergistic parts:
• Alexandria is the world’s largest commercially-usable audio dataset for machine learning, featuring millions of audio files that are structured taxonomically with full data provenance.
• AuditoryNET is a highly-optimised deep neural network for sound recognition, which models the ideophonic features of sounds.
In addition to core technology are nearly 50 experts in acoustics, data, machine learning and embedded software engineering.
The company has successfully licensed its technology to major global brands, including two of the world’s biggest companies, Hive, Iliad, Sengled and others. The company has partners including Arm, Intel, Knowles, Ambarella, Ambiq, Vesper, Frontier and others.
Contact Details and Archive...