Do you have any suggestions or assistance in using the speech to text function to recognize homophones that may cause errors.

连博10335043 65

like for Chinese "枯(Ku)"，recognized as "哭(Ku)".Cannot contact context.

This is just a probabilistic issue.

dupammi 8,615 Reputation points Microsoft External Staff

2024-07-29T09:54:42.79+00:00

Hi @连博10335043

Thank you for your question.

To improve homophone recognition in speech-to-text, try providing clear context in your speech and using any available custom vocabularies or lexicons. Ensure proper punctuation and formatting to reduce ambiguity. Providing feedback on incorrect transcriptions can also help enhance accuracy. For persistent issues, consider reaching out to Azure support for further assistance.
连博10335043 65 Reputation points

2024-07-29T11:34:59.2366667+00:00

Hi,What's the "custom vocabularies or lexicons",my understanding is the personal voice for specific vocabulary.

If that's the case, is it necessary to collect voice data in this area for training to improve recognition rate.

Thank you.
dupammi 8,615 Reputation points Microsoft External Staff

2024-07-31T05:35:42.4333333+00:00

Hi @连博10335043

To enhance homophone recognition in speech-to-text, use lexicons to define specific terms and their pronunciations. Providing clear context and proper grammar in your speech helps reduce ambiguity. Collecting and training with relevant voice data can improve accuracy. Regularly report transcription errors to improve system learning. For advanced needs, consider Azure’s Custom Speech model for tailored training. If issues persist, contact Azure support for further assistance.

Your answer