Speech-to-text: Disfluency Removal configuration

KEN KIM 16

I am using the speech-to-text REST API (python) to do some research regarding fillers, pauses, and backtracking in Japanese (ja-JP).

Can I config disfluency removal while using the Speech-to-text service? I need to have true text with all the fillers in the transcribed text.

Thank you in advance.

YutongTie-MSFT 51,501 Reputation points

2022-12-17T12:20:52.967+00:00

Hello @KEN KIM

Thanks for reaching out to us, what do you mean by configing? What you want to config? More information pls refer to - https://learn.microsoft.com/en-us/azure/cognitive-services/speech-service/display-text-format?pivots=programming-language-java#disfluency-removal

Regards,
Yutong
KEN KIM 16 Reputation points

2022-12-22T13:06:09.32+00:00

Hi,

Azure STT service currently automatically removes filler words, but I need to have those filler words such as "eto" or "ano" after I do STT.

Are there any options to allow me to include those filler words when displaying text?

Thank you.
YutongTie-MSFT 51,501 Reputation points

2022-12-27T15:02:11.633+00:00

Hello @KEN KIM

I can understand your scenario, I am checking with the pm to comfirm the way to do that, will let you know soon. Thanks for your understanding.

Regards,
Yutong
KEN KIM 16 Reputation points

2023-01-17T11:05:07.44+00:00

Hi @YutongTie-MSFT ,

Please let me know if you have any updates.

Thank you for the help.

Best Regards,

Kon Woo Kim
Deleted

This comment has been deleted due to a violation of our Code of Conduct. The comment was manually reported or identified through automated detection before action was taken. Please refer to our Code of Conduct for more information.
KEN KIM 16 Reputation points

2023-01-21T11:14:19.1833333+00:00

Hi @YutongTie-MSFT ,

Please let me know if you have any updates.

Best Regards,

Kon Woo Kim
Jen Yu 0 Reputation points Microsoft Employee

2023-03-16T07:44:12.01+00:00

Hi @YutongTie-MSFT ,

I am also curious about this question. Any updates on how can we config to remove words like "eto" or "ano"?
Devanshi Thakar 20 Reputation points Microsoft Employee

2024-02-08T22:27:50.4366667+00:00

Curious if there are any updates on this, thank you! @YutongTie-MSFT
hotrod 1 Reputation point

2024-10-02T20:22:50.1766667+00:00

Comment edited, see below:
hotrod 1 Reputation point

2024-10-02T20:43:28.2166667+00:00
Hello @ken kim @YutongTie-MSFT

Same issue, would like to disable Disfluency Removal when using the Azure Cognitive Services Speech SDK for Python. From https://learn.microsoft.com/en-us/answers/questions/1165008/how-to-config-disfluency-removal-using-rest-api?page=1 :

"I have done some research for a similar query before, and it seems that a property SpeechServiceResponse_PostProcessingOption needs to be set through SDK to stop this feature or automatic disfluency removal".

I cannot find any reference to set this property in the Azure Cognitive Services Speech SDK for Python. Expect it to be something like this:

properties = swagger_client.TranscriptionProperties() properties.SpeechServiceResponse_PostProcessingOption = "None"

Share via

Speech-to-text: Disfluency Removal configuration

Your answer