Displaying 1 to 6 from 6 results

Porcupine - On-device wake word detection engine powered by deep learning.

  •    C

Try out Porcupine using its interactive web demo. You need a working microphone. Try out Porcupine by downloading it's Android demo application. The demo application allows you to test Porcupine on a variety of wake words in any environment.

sonus - :speech_balloon: /so.nus/ STT (speech to text) for Node with offline hotword detection

  •    Javascript

Sonus lets you quickly and easily add a VUI (Voice User Interface) to any hardware or software project. Just like Alexa, Google Now, and Siri, Sonus is always listening offline for a customizable hotword. Once that hotword is detected your speech is streamed to the cloud recognition service of your choice - then you get the results. Generally, running npm install should suffice. This module however, requires you to install SoX.

sonus-electron-boilerplate - Example of sonus running in electron

  •    Javascript

Example of Sonus running in electron for OSX and Linux. What you get: customizable offline hotword detection and streaming recognition for your stand-alone desktop application.

mycroft-precise - A lightweight, simple-to-use, RNN wake word listener

  •    Python

A lightweight, simple-to-use, RNN wake word listener. Precise is a wake word listener. Like its name suggests, a wake word listener's job is to continually listen to sounds and speech around the device, and activate when the sounds or speech match a wake word. Unlike other machine learning hotword detection tools, Mycroft Precise is fully open source. Take a look at a comparison here.

go-snowboy - Go wrapper for Kitt-AI's snowboy audio detection library.

  •    Go

The Go bindings for snowboy audio detection (https://github.com/Kitt-AI/snowboy) are generated using swig which creates a lot of extra types and uses calls with variable arguments. This makes writing integrations in golang difficult because the types aren't explicit. go-snowboy is intended to be a wrapper around the swig-generated Go code which will provide Go-style usage. Example hotword detection usage in example/detect.go. Example API hotword training usage in example/api.go.

wakeword-benchmark - wake word engine benchmark framework

  •    Python

The primary purpose of this benchmark framework is to provide a scientific comparison between different wake-word detection engines in terms of accuracy and runtime metrics. Currently, the framework is configured for Alexa as the test wake-word. But it can be configured for any other wake-words as described here. Common Voice is used as background dataset, i.e., dataset without utterances of the wake-word. It can be downloaded from here. Only recordings with at least two up-votes and no down-votes are used (this reduces the size of the dataset to ~125 hours).