Azure cognitive speech Produces missing/incorrect words

Question

Hi, I am wondering why the extracted text from audio have missing/incorrect words.

In this case the language used is Greek(el-gr)(other language could have the same issue)

Below are the image of Transcriptions JSON result and the human edited/correct captions.

Correct words are those that are in black background.

and here is the payload request for creating a transcription.

 var azureTranscription = new AzureTranscription
 {
     ContentUrls = new List<Uri> { contentUri },
     Locale = locale,
     DisplayName = displayName,
     Properties = new AzureTranscriptionProperties
     {
         Channels = new List<int> { 0 },
         WordLevelTimestampsEnabled = true,
         PunctuationMode = punctuationMode,
         ProfanityFilterMode = profanityFilterMode,
         TimeToLive = TimeSpan.FromHours(24),
         DestinationContainerUrl = destinationContainerUrl,
     },
     
 };

Are there special settings I need to consider for creating Transcription for languages like Greek?

Thanks