Skip to content

Commit 7623fc0

Browse files
jimxieieric-urban
andauthored
Update articles/cognitive-services/Speech-Service/includes/how-to/conversation-transcription/real-time-javascript.md
Co-authored-by: Eric Urban <[email protected]>
1 parent 70e0f04 commit 7623fc0

File tree

1 file changed

+1
-2
lines changed

1 file changed

+1
-2
lines changed

articles/cognitive-services/Speech-Service/includes/how-to/conversation-transcription/real-time-javascript.md

Lines changed: 1 addition & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -74,8 +74,7 @@ This sample code does the following:
7474
* Registers to events and begins transcription.
7575
* If you want to differentiate speakers without providing voice samples, please enable `DifferentiateGuestSpeakers` feature as in [Conversation Transcription Overview](../../../conversation-transcription.md).
7676

77-
> [!NOTE]
78-
> If speaker identification or differentiate is enabled, even you have received `transcribed` results, service is still evaluating them by accumulated audio information. When service finds any previous result assigned a wrong `speakerId`, it will send out an almost same `transcribed` result again, but only the `speakerId` and `UtteranceId` are different. Since the `UtteranceId` format is `{index}_{speakerId}_{Offset}`, when you receive a `transcribed` result, you could use `UtteranceId` to determine if current `transcribed` result is going to correct previous one. Your client or UI logic could decide behaviors, like overwriting previous output, or just ignore the current one.
77+
If speaker identification or differentiate is enabled, then even if you have already received `transcribed` results, the service is still evaluating them by accumulated audio information. If the service finds that any previous result was assigned an incorrect `speakerId`, then a nearly identical `Transcribed` result will be sent again, where only the `speakerId` and `UtteranceId` are different. Since the `UtteranceId` format is `{index}_{speakerId}_{Offset}`, when you receive a `transcribed` result, you could use `UtteranceId` to determine if the current `transcribed` result is going to correct a previous one. Your client or UI logic could decide behaviors, like overwriting previous output, or to ignore the latest result.
7978

8079
```javascript
8180
(function() {

0 commit comments

Comments
 (0)