Do we have a WebSocket API reference for continuous Speech Recognition?

[parijat1209]

Hi,

Speech.JS cannot do continuous recognition as the Partial Response is never received. I believe SpeechClient dll is connecting to Bing WebSocket servers for the continuous recognition to happen. Can't we directly connect to Bing WebSocket servers from HTML5 WebSockets? If so, then where can I find a sample code or a reference to directly connect to the Bing Speech WebSocket servers using JavaScript WebSockets?

Thanks

[dmiller235623]

connect to my outlook email not my gmail.. I get a lot of
spam...dmiller235623@outlook.com

On Mon, Apr 25, 2016 at 3:33 AM, parijat1209 notifications@github.com
wrote:

Hi,

Speech.JS cannot do continuous recognition as the Partial Response is
never received. I believe SpeechClient dll is connecting to Bing WebSocket
servers for the continuous recognition to happen. Can't we directly connect
to Bing WebSocket servers from HTML5 WebSockets? If so, then where can I
find a sample code or a reference to directly connect to the Bing Speech
WebSocket servers using JavaScript WebSockets?

Thanks

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub
#58

[ochakov]

+1

[sushmaakoju]

We are looking at disable Barge-in on ASR (c#.net) module of Project Oxford Speech API. Can you please help suggest us onm this?

[margaretmz]

This issue was moved to Microsoft/Cognitive-Speech-STT-JavaScript#4

[willesandstrom]

I'm also looking for this. I've looked at the request from the client SDK for Windows using Fiddler and there is web socket endpoints available there

This are what the sent headers from the library looks like

https://websockets.platform.bing.com/ws/speech/recognize

GET /ws/speech/recognize HTTP/1.1
Connection: Upgrade
Upgrade: Websocket
Authorization: Bearer ashdjashdjlasd
User-Agent: Mozilla/4.0 (Windows 8; Unknown;Unknown;ProcessName/AppName=Unknown;DeviceType=Near;SpeechClient=1.0.716436)
X-WebSocketConnectionId: 4eb71213ef974307826d4c89639ddf53
X-CU-LogLevel: 1
X-Search-AppID: Unknown
X-Search-Market: en-US
X-Search-UILang: en-US
Sec-WebSocket-Key: lS4KX2vWHDnmiO3UnECc8A==
Sec-WebSocket-Version: 13
Host: websockets.platform.bing.com

This makes things bad since I cannot add custom headers to a websocket in JS.

If you are able to add the authorization header to your websocket request, then you can use the code above. Also see their demo which uses partial response/streaming, but it uses it's own endpoint (https://cog-web-wu.azurewebsites.net/cognitive-services/ws/speechtotextdemo?language=en-US&g_Recaptcha_Response=null&isNeedVerify=false). If we could somehow add the authorization header to a web socket, we could replace the demo endpoint with the real endpoint and have a full working partial response support in JS.