Speech to text - Norwegian: Capitalization

Gunnar Sylthe 6 Reputation points
2021-03-15T09:27:32.823+00:00

We're using Microsoft.CognitiveServices.Speech for transcription/subtitling of video clips, mostly Norwegian materials. We have noticed that e.g. spelling of proper nouns is impressively correct, but that capitalization has been missing. But as of this weekend, there seems to be a change. Now, there is suddenly TOO MUCH capitalization going on. E.g., all occurrences of the word "nok" is written in all caps (which makes it look like the abbreviation for Norwegian currency (NOK)). The same thing happens for certain other words, like "FRA" and "ET". Also, seemingly random words in the middle of sentences are capitalized. Is this a bug MS is aware of, so that we can expect a fix soon?

C#
C#
An object-oriented and type-safe programming language that has its roots in the C family of languages and includes support for component-oriented programming.
10,627 questions
Azure AI services
Azure AI services
A group of Azure services, SDKs, and APIs designed to make apps more intelligent, engaging, and discoverable.
2,627 questions
{count} votes

1 answer

Sort by: Most helpful
  1. GiftA-MSFT 11,161 Reputation points
    2021-03-19T20:59:06.443+00:00

    A fix has been rolled out, issue should be resolved now, let us know if otherwise. Sorry for the inconvenience. Thanks.