IntentRecognizer not working with LUIS WestEurope app
See original GitHub issueDescribe the bug
I tried to use IntentRecognizer
with a LUIS app, hosted in WestEurope (https://eu.luis.ai).
I started with the samples and was quickly blocked with this IntentRecognizer
, where when I send an audio (from audio file for example), the text is well transcribed but then nothing happened about recognition. I just got a canceled event after a long time:
Say something...
Session started event.
RECOGNIZING: Text=tell
RECOGNIZING: Text=current
RECOGNIZING: Text=current weather
CANCELED: Reason=Error
CANCELED: ErrorCode=ConnectionFailure
CANCELED: ErrorDetails=WebSocket operation failed. Internal error: 3. Error details: WS_ERROR_UNDERLYING_IO_ERROR
CANCELED: Did you update the subscription info?
Recognition done.
I used exactly the same application on LUIS West US portal (https://www.luis.ai), changing LUIS Subscription key, LUIS App Id and LUIS region, and it is working well:
Say something...
Session started event.
RECOGNIZING: Text=tell
RECOGNIZING: Text=current
RECOGNIZING: Text=current weather
RECOGNIZED: Text=Current weather.
Intent Id: Weather.GetCondition.
Language Understanding JSON: {
"query": "current weather",
"topScoringIntent": {
"intent": "Weather.GetCondition",
"score": 0.9494868
},
"entities": []
}.
NOMATCH: Speech could not be recognized.
NOMATCH: Speech could not be recognized.
CANCELED: Reason=EndOfStream
Session stopped event.
Stop recognition.
Recognition done.
In both cases, I used subscription keys of type S0 (not free), apps are targeting English and Speech is enabled in LUIS settings.
Capture (1st call: West US, 2nd call: West Europe):
To Reproduce Steps to reproduce the behavior:
- Create an app in LUIS WestEurope, which language will be English
- Add a pre-built domain intent for example, like Weather.GetCondition
- Add Speech capability in LUIS settings
- Train and publish.
- Use
IntentRecognitionSamples
in this repository
Expected behavior IntentRecognizer should return the same thing between WestUs and WestEurope
Version of the Cognitive Services Speech SDK Latest Microsoft.CognitiveServices.Speech NuGet package (1.2.0) for C#
Platform, Operationg System, and Programming Language
- OS: Windows 10 (VM Parallels on Mac)
- Hardware - x64
Issue Analytics
- State:
- Created 5 years ago
- Comments:11 (5 by maintainers)
Top GitHub Comments
Great, I will check tonight once I have a moment and give you a feedback
This should be fixed from the service side this week. I’ll give an update once the fix is deployed. Thanks for your patience,
Mark