Cross-browser Voice AI with Picovoice Web SDKs now Available

Picovoice has released its new Web SDK with out-of-the-box support across all major frameworks, including React, Vue, and Angular. With the introduction of voice AI SDKs for the web, Picovoice has become the first and only end-to-end voice recognition platform for web browsers.

The Picovoice web packages work on all modern browsers. Picovoice runs voice AI privately inside the browser without transmitting audio data to the cloud. Unlike the SpeechRecognition API that only works in Google Chrome, Picovoice works on Chrome, Firefox, Edge, Safari, and Electron.

Developer-focused

Picovoice engines for the web are now available as npm packages with both modern and legacy JavaScript support. Developers can get on-device voice AI working in a web application in about ten lines of code. React, Vue, and Angular frameworks receive dedicated packages for ease of integration, including microphone access and audio preprocessing. TypeScript support and JavaScript modules enable best practices and also help developers do their job with imports, code autocomplete, and type checking.

Cutting-edge technology abstracted into a simple API

Behind the scenes, the web packages operate using WebAssembly, the Web Audio API, and Web Workers, keeping your application running smoothly by avoiding congesting the main JavaScript thread. The same extreme efficiency that allows Picovoice to run on microcontrollers lends itself to minimal resource usage on the web. These details are all abstracted by the SDK, leaving a straightforward API.

Demo

Here is a demo showing the Web SDKs. It is running English, German, Spanish, French, Italian, Japanese, Korean and Portuguese Porcupine Wake Word engines simultaneously:

Start Developing Today