Merge pull request #101505 from IEvangelist/speechContainersFAQ

Initial bits of FAQ page for speech

Author: ttorble

articles/cognitive-services/Speech-Service/includes/speech-modes.md

@@ -0,0 +1,32 @@
+---
+author: IEvangelist
+manager: nitinme
+ms.service: cognitive-services
+ms.topic: include
+ms.date: 01/22/2020
+ms.author: dapine
+---
+
+## Speech modes
+
+**Interactive**
+- Meant for command and control scenarios.
+- Has a segmentation time out value of X.
+- At the end of one recognized utterance, the service stops processing audio from that request ID and ends the turn. The connection is not closed.
+- Maximum limit for recognition is 20s.
+- Typical Carbon call to invoke is `RecognizeOnceAsync`.
+
+**Conversation**
+- Meant for longer running recognitions.
+- Has a segmentation time out value of Y. (Y != X)
+- Will process multiple complete utterances without ending the turn.
+- Will end the turn for too much silence.
+- Carbon will continue with a new request ID and replaying audio as needed.
+- The service will forcibly disconnect after 10 minutes of speech recognition.
+- Carbon will reconnect and replay unacknowledged audio.
+- Invoked in Carbon with `StartContinuousRecognition`.
+
+**Dictation**
+- Allows users to specify punctuation by speaking it.
+- Invoked in Carbon by specifying `EnableDictation` on the `SpeechConfig` object regardless of the API call that starts recognition.
+- The 1<sup>st</sup> party cluster returns `speech.fragment` messages for intermediate results, the 3<sup>rd</sup> party return `speech.hypothesis` messages.

articles/cognitive-services/Speech-Service/index-hosting.yml

@@ -10,7 +10,7 @@ metadata:
   manager: nitinme
   ms.service: speech-service
   ms.topic: landing-page
-  ms.date: 03/10/2020
+  ms.date: 03/23/2020
   ms.author: dapine
 
 landingContent:
@@ -40,3 +40,5 @@ landingContent:
           url: regions.md
         - text: Sovereign clouds
           url: sovereign-clouds.md
+        - text: Containers FAQ
+          url: speech-container-faq.md