How to identify filler words in Azure AI Speech

Question

How to identify filler words in Azure AI Speech

Sai Vishnu Soudri 65

Hi team.

Is there any feature in Azure Speech that can help us identify filler words?

Please point me to the right documentation if there is any.

Thanks,

Sai Vishnu Soudri

0 comments

Answer accepted by question author

1 additional answer

Your answer

Answer 1

SRILAKSHMI C 16,975 Microsoft External Staff Moderator

Hello Sai Vishnu Soudri,

Welcome to Microsoft Q&A! Thanks for posting the question.

Azure AI Speech has a feature called Disfluency Removal that can help identify and remove filler words from transcriptions. This feature is particularly useful for creating cleaner, more readable text from spoken language.

To enable or disable this feature, you can use the TrueText option in the Speech SDK. When TrueText is enabled, the service will automatically remove disfluencies from the transcribed text.

For more information, you can refer to the official documentation, Disfluency Removal.

Hope this helps. Do let us know if you any further queries.

If this answers your query, do click Accept Answer and Yes for was this answer helpful.

Thank you!

SRILAKSHMI C 16,975 Reputation points Microsoft External Staff Moderator

2024-11-12T11:08:28.3433333+00:00

Hi Sai Vishnu Soudri,

Following up to see if the above answer was helpful. If this answers your query, do click Accept Answer and Yes for was this answer helpful. And, if you have any further query do let us know.
Sai Vishnu Soudri 65 Reputation points

2024-11-13T09:29:42.3533333+00:00

Thank you for the response. I have tried the disfluency removal feature and it’s quite handy.

I did notice that if I just say uhh or hmm, it sometimes gets transcribed. To catch these, does Azure speech have any mechanism of always returning the same spelling for certain filler words?

Eg: “uhhhhh” is always returned as “uh” and “Hmmmmmm” is always returned as “Hmm”.
SRILAKSHMI C 16,975 Reputation points Microsoft External Staff Moderator

2024-11-13T12:06:30.1433333+00:00

Hi Sai Vishnu Soudri,

Azure Speech services doesn’t have a built-in mechanism to directly standardize varying spellings of filler sounds into a single version. However, applying a post-processing script can further refine this, ensuring that words like “uhhhh” and “hmmmm” are shortened to “uh” or “hmm” as needed.

Thank you!
SRILAKSHMI C 16,975 Reputation points Microsoft External Staff Moderator

2024-11-14T11:26:34.4433333+00:00

Hi Sai Vishnu Soudri,

Just checking in to see if the below answer helped. If this answers your query, do click Accept Answer and Yes for was this answer helpful. And, if you have any further query do let us know. Thank you!

Answer 2

Abiola Akinbade 30,490 Volunteer Moderator

Maybe disable disfluency removal, you can obtain transcriptions that include filler words
See: https://learn.microsoft.com/en-us/azure/ai-services/speech-service/display-text-format?pivots=programming-language-csharp

You can mark it 'Accept Answer' and 'Upvote' if this helped you

Regards,

Abiola

Sai Vishnu Soudri 65 Reputation points

2024-11-13T09:28:42.0633333+00:00

Thank you for the response! I have tried the disfluency removal feature and it’s quite handy.

I did notice that if I just say uhh or hmm, it sometimes gets transcribed. To catch these, does Azure speech have any mechanism of always returning the same spelling for certain filler words?

Eg: “uhhhhh” is always returned as “uh” or “Hmmmmmm” is always returned as “Hmm”. Thanks,

Sai Vishnu Soudri

Share via

How to identify filler words in Azure AI Speech

1 additional answer

Your answer