Azure-docs: How do you overcome infinite voice cycling...

[Nesyne]

[Enter feedback here]
I am currently doing Speech-to-Speech translation. My issue is, if i am using my laptops mic and speaker, my mic is picking the voice from speaker and trying to re-translate that; which gets into an infinite re-translation.

---

Document Details

⚠ Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.

• ID: c71b1b04-27c4-8b1f-2de1-f2205628cffc
• Version Independent ID: 1ea51642-e48e-57b3-b9c4-7a3a661a4f5d
• Content: Quickstart: Translate speech-to-speech - Speech service - Azure Cognitive Services
• Content Source: articles/cognitive-services/Speech-Service/quickstarts/translate-speech-to-speech.md
• Service: cognitive-services
• Sub-service: speech-service
• GitHub Login: @IEvangelist
• Microsoft Alias: dapine

[RohitMungi-MSFT]

@Nesyne Could you please let us know if you are using any specific quickstart doc or link when this behavior is seen to replicate the scenario?

[robch]

This isn't specific to Azure Speech SDK/Service... In the field of speech recognition, the best solution to this problem is called "echo cancellation".

On Windows, some hardware has advanced audio processing techniques that will effectively "remove the audio output" from the "audio input". You can read a bit more about that here.

The Azure Speech SDK will attempt to open the audio device on Windows by requesting to use the Speech audio input category (see here for a bit more on that).

Similar approaches are used on other operating system platforms when available. Is this an issue you're experiencing on Windows or another platform?

[IEvangelist]

With the details from @robch here, I'm inclined to mark this issue as resolved, but I'd love to ensure that the customer is satisfied with the response. Please let us know @Nesyne if this is helpful and if we can close this issue.

Thank you

[Nesyne]

Thanks for your explanation @robch. I will go ahead and mark this issue as solved.