Voice-Enabled Web: On-device Voice Recognition for Browsers

WebSpeech API enabled websites to offer voice experiences. Google Translate and Google Search by Voice Engine are some examples of WebSpeechAPI applications. However, WebSpeech API has some challenges and limitations. It does not offer always-on listening, i.e. truly hands-free voice experience or offline voice recognition i.e. private and cost-effective voice recognition with no latency cross-browser support. For all these reasons and more voice adoption for the web has been limited, despite its promise to make life easier for everyone. Picovoice overcomes these challenges to help enterprises improve individuals’ experiences.

We pay our credit card bills every month, yet most of us struggle to find the payment portal in the banking web application. Filling complex forms, setting filters, and navigating a new website are also onerous tasks that web users have to perform. Small screens make these tasks even more challenging. Users are intimidated when they see a long page with forms and selections, in fact, 23% of them abandon a checkout [1]. The attention of a user is scarce, they form an opinion about a website in 0.05 seconds, and 55% of them leave the website in 15 seconds [2] [3].

A website visitor either intends to search for information or to perform a task, regardless of the nature of the page. Voice interfaces elevate user experience in both scenarios. A shopper can ask for women size 10 black Nike running shoes with a rating above 4 stars instead of finding and setting filters one-by-one on a mobile device. She can also say “add to cart” instead of scrolling to find it. Adding commands such as switching to dark mode, reading more, helps users achieve what they want promptly. A salesperson can be up to 40% more effective when using a voice-enabled CRM, than typing.

To add voice to a website, if restricting users to Chrome or newer Safari is acceptable, one may decide to rely on Web Speech API for free [4]. If not, AWS, Google, Azure, and IBM offer cross-browser cloud services but can be unbearably costly at scale. Plus, none of these approaches are feasible to provide a hands-free experience and require sending users’ voice data to third-party servers. Picovoice cross-browser SDK is unique as it offers a hands-free web experience across browsers without risking your users’ trust or inflating your cloud bill.

Cross-browser support: Using Picovoice cross-browser SDK, you design and develop once and deploy into Chrome, Safari, Firefox, Edge, Opera, etc.

Check out Picovoice Web SDKs for cross-browser voice recognition

Hands-free experience: Picovoice Porcupine is the only option if you want wake word or always-listening voice commands for your web applications. Push-to-talk solutions hinder the user experience, especially on mobile devices. They take up to 30% of mobile screens. Cost-effective at scale: Cloud providers charge per API call. Google DialogFlow costs $0.06 for sessions up to a minute [5]. Given the average session duration is 10 minutes on Amazon [6], every visitor will cost $0.6 and 100,000 visitors per day will turn into a daily bill of $60,000. Picovoice offers unlimited interactions instead of limiting with an API call, hence no excessive bills.

Cost-effective at scale: Cloud providers charge per API call. Google DialogFlow costs $0.06 for sessions up to a minute. 5 Given the average session lasts 10 minutes on Amazon 6, every visitor will cost $0.6 and 100,000 visitors per day will turn into a daily bill of $60,000. Picovoice offers unlimited interactions instead of limiting with an API call, hence no excessive bills.

Try now!

To experience how voice works on a browser, try this demo. Even better? Turn off your internet connection and see that it’s working offline.

Start Building!

If you’re interested in developing your own voice product, try Picovoice Console for free and build your prototype in minutes with Picovoice SDKs for Angular, React, and Vue.

Learn more!

If you’re happy with the demo and excited about the things that you can offer with on-device voice processing, start for free or get expert help.