Introducing KeenASR for Web

Unlock the Power of On-Device Speech Recognition in Your Web Applications

Author: Ognjen Todic | April 16, 2023

At Keen Research, we have been developing on-device speech recognition solutions for nearly seven years now. We started with supporting the iOS operating system, but soon realized the need for an Android version of our SDK. After years of providing support for both platforms, we are excited to announce the launch of KeenASR SDK for Web, a JavaScript library that performs speech recognition locally (“offline”) in the browser, just like its iOS and Android counterparts.

Unlike cloud-based systems KeenASR for Web ensures speech data is processed directly in the browser, giving you complete control over user privacy, scalability, and customization. This approach also provides a seamless, responsive user experience without the hiccups that can arise from spotty internet connectivity.

Why Choose On-Device Processing?

The on-device approach provides several benefits over cloud-based speech recognition solutions. Firstly, it ensures user privacy by not requiring the transfer of speech data to the cloud. Secondly, it provides a more robust user experience, especially for long listening sessions, as even short hiccups in internet connection can be detrimental. Scalability is not an issue either, since all processing is done locally, eliminating the need for additional server provisioning as usage grows. With predictable costs that do not depend on usage or number of users, on-device processing is highly cost-effective at scale. The highly customizable KeenASR SDK also allows for personalization of user experience, which is not always possible with cloud-based solutions. Moreover, there is no latency due to round-trip to the server, making it ideal for use cases where low latency is important.

Web Applications and Offline Mode

What’s more, with HTML5, voice web applications can now be run in offline mode as well. KeenASR SDK for offline speech recognition already powers a number of iOS and Android mobile apps focused on education, augmented intelligence for frontline workers, helping people with disabilities, and many other use cases. With web application support, we see a number of new opportunities for EdTech and other companies to leverage our support for speech recognition optimized for kids’ voices and use cases such as oral reading and assessment, language learning, and early education.

Beta Release and Future Plans

Currently in closed-beta, we plan to release a beta version of the SDK in May 2023. During the open beta release, developers will be able to download and test the trial JavasScript library from our website, which will include several demo web applications showcasing the use of KeenASR SDK. And stay tuned for our upcoming release of web support in the KeenASR Unity plugin later this year.

Join Us in Shaping the Future of On-Device Speech Recognition

Contact us today to be a part of our closed-beta program and try out our SDK for your web application needs. Or, checkout our open positions if you are interested in joining a journey.

We provide professional services for SDK integration, proof-of-concept development, customization of language and acoustic models, and porting to custom hardware platforms.