/mmir-plugin-asr-nuance-ws

mmir plugin for Nuance WebSockets service (MIX) for speech recognition

Primary LanguageJavaScriptMIT LicenseMIT

mmir-plugin-asr-nuance-ws

DEPRECATED this plugin is outdated since Nuance SpeechKit service is discontinued, for a current alternative see mmir-plugin-asr-cerence-ws 

- DEPRECATED Nuance SpeechKit service is discontinued
- for current alternative see [mmir-plugin-asr-cerence-ws](https://github.com/mmig/mmir-plugin-asr-cerence-ws) based on the Cerence WebSocket API

Cordova plugin for the MMIR framework that allows Automatic Speech Recognition (ASR) via Nuance WebSockets services (MIX)

configure CSP

(e.g. index.html): allow access to https://dictation.nuancemobility.net

  <meta http-equiv="Content-Security-Policy"
        content="default-src 'self' 'unsafe-inline' 'unsafe-eval' https://dictation.nuancemobility.net ...

configuration.json:

{

...

	"asrNuanceWs": {
		"encoder": "speexEncoder",
		"appId": <the app ID>,
		"appKey": <the secret app key>
	},

	....

	"mediaManager": {
    	"plugins": {
    		"browser": [
    			...
                {"ctx": "webn", "mod": "webAudioInput", "config": "asrNuanceWs"}
                ...
    		],
    		"cordova": [
    			...
                {"ctx": "webn", "mod": "webAudioInput", "config": "asrNuanceWs"}
                ...
    		]
    	}
    },
...

}

options

supported options for recoginze() / startRecord():

  • language: String
  • results: Number
  • mode: 'search' | 'dictation'

supported custom options for recoginze() / startRecord():

  • appKey: String
  • appId: String
  • codec: 'speex' | 'wav'
  • source: "SpeakerAndMicrophone" | "HeadsetInOut" | "HeadsetBT" | "HeadPhone" | "LineOut"
    source: Indicates the source of the audio recording.
    Properly specifying this header improves recognition accuracy.
    Nuance encourages you to pass this header whenever you can -- and as accurately as possible.