Speech SDK speech to text segmentation silence timeout is not working as intended

H M Moniruzzaman 0 Reputation points
2024-12-10T11:27:34.7566667+00:00

I am using Azure Translation recognizer to recognize and translate text. I am also using the segmentation silence timeout to detect shorter pauses and break sentences as short as possible.

According to documentation the supported values are between 100 to 5000 ms. But when I set this to 100ms to 500ms, this seems to be not working (not breaking the sentences at these timeouts). For example when I set it to 100ms it still cannot detect 500ms of pauses. Here is how I am setting it:

speechConfig.setProperty(PropertyId.Speech_SegmentationSilenceTimeoutMs, '100');

And this is how I am using the TranslationRecognizer:

this.recognizer = new speechsdk.TranslationRecognizer(speechConfig, audioConfig);

this.recognizer.startContinuousRecognitionAsync();


Any idea how to resolve this? The goal is to get really short segmented text blocks. Or is there any other relevant parameters to achieve this?

Thank you in advance for your answer.

Azure AI Speech
Azure AI Speech
An Azure service that integrates speech processing into apps and services.
1,835 questions
{count} votes

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.