-
Notifications
You must be signed in to change notification settings - Fork 1.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Cognitive Speech not working in West Europe #2802
Comments
Hi @wehnicde, I have been unable to repro your issue. I have a couple clarifying questions for you:
|
I am running in Chrome but I need IE support too, thats why I use es5. (Does it have any disadvantages?)
It isn't. The "taskbar" on top is simple html by me (using style.display = none), it is the standart webchat.
Didn't help :( When I change to 4.5.3 the webchat won't render with the error:
If I take the speech stuff out, it works. With 4.6 it is rendering, did 4.6 change anything about the implementation of webSpeech? This is my complete HTML-Code (clean version).
Didn't change anything.
I will try to generate a fresh bot and try to repro it by myself again. My problem is, that I already have 2 seperate bots (C# and JS) and it isn't working on both, so I really don't know what the problem could be. Edit: Just generated a clean bot and put the secret in my html. Didn't work. So the bot-code shouldn't be the problem. |
@wehnicde, I'm a little confused on whether you are trying to get this to work in Chrome, IE, or both. If you are having issues with Chrome, I'm certain we can get it working. Your implementation is slightly different but really should work. And, if not, we can play with the code. As for IE 11, as stated here, Internet Explorer does not meet the minimum requirements for supporting speech recognition and speech synthesis, at least as far as Cognitive Services Speech is concerned. |
Sorry for the confusing. I want to have the webchat working on all browsers, but I know that speech isn't possible with IE11. Writing with the bot should work on every browser, speech only with Chrome/Firefox. Thats why I need to implement the es5 file (Am I right about that?) What do you mean by "Your implementation is slightly different but really should work"? How is my implementation different? And how is the "normal" way? Maybe it will work with that. |
I am getting the same problem in IE, Chrome, Edge Chrome. I should note that the bot still works in Chrome for typing, when you press the microphone for speech it does ask to allow, then the errors start pouring in and no actual listening seems to happen. |
Adding my voice: getting same error in Chrome, a couple days ago Speech was working, now without changing anything in the code I'm getting webchat.js:2 WebSocket is already in CLOSING or CLOSED state. I noticed this issue was just closed (10 hours ago): anything to do with it maybe? (regression?) |
@wehnicde, yes, the es5 version will work in all web browsers the same. Outside of the file package size if that is a concern, you should be just fine. Never mind my comment about implementation. You're is just fine and acceptable. I thought I saw you doing something slightly different, but I was mistaken. I did go ahead and create a new CS Speech service a second time (and a new bot for a clean slate tho I also confirmed with an established bot) and was able to repro. I will have an update when we know more. |
Reassigning @stevkan for azure reporting only |
This bug is very similar to Azure-Samples/cognitive-services-speech-sdk#127. West Europe services is returning "Specified grammar type is not supported!" when we pass LUIS reference grammar ID. When we take away LUIS reference grammar ID (in hacked code), it works with West Europe. So I am assuming Cognitive Services in West Europe is having issue on handling LUIS reference grammar ID. @stevkan could you file a ticket to them, mentioning LUIS reference grammar ID (a.k.a. IntentRecognizer) is not working West Europe? I have pinged @mahilleb-msft offline, but he is in Munich time zone, no response yet. |
@compulim I am not using LUIS for my bot. Or does Cognitve Speech use LUIS by itself? |
Although you may not using LUIS in your bot, Direct Line channel do provides a LUIS ID and requires Web Chat to pass it to Cognitive Services. The ID is called "LUIS reference grammar ID". Even in your case, you are not using LUIS, Direct Line channel still send us the LUIS ID, so we still pass it to Cognitive Services. And I believe Cognitive Services has some issues handling LUIS ID in West Europe region. We have contacted Cognitive Services team and they are investigating the issue. |
Oh okay. Thanks for the clarification. And thanks for the investigation. |
@stevkan The error seems different today. Not sure if you changed anything, I did not: (It's similar) |
Web Chat only own the client-side. I believe the service team is working on a fix. |
I'm receiving a similar issue but my Speech is set to the UK South region. Would that also be affected by this? |
Hi, I am getting this same WebSocket issue in locale:en-US and speech region: eastus Is there any fix for this? We are not able to use speech service because of this. |
Any news about the "WebSocket is already in CLOSING or CLOSED state." issue? I tried with northeurope and it seems to fail too. |
Closing this issue as it is working now. Thanks for the investigation. |
Thanks @wehnicde for reporting the problem. |
Screenshots
Version
I use 4.7.1, but it doesn't work on 4.6.0 too.
Describe the bug
When I click on the microphone this error occurs.
On 19.12.2019 everything worked fine and when I tried on 20.12.2019 WITHOUT changing anything at the code it wouldn't work.
This is how I implement the Cognitive Speech Service:
Expected behavior
It should recognize my voice and translate it to text.
Additional context
My bot runs on C#, but I tried on a JS-Bot too, same Error. WebSockets are enabled in the BotConfig.
[Bug]
The text was updated successfully, but these errors were encountered: