@1283 更科 明 You can lookup this sample from the speech SDK repo to use the helper class.
The method RecognitionWithPullAudioStreamAsync() uses the helper class to read the input file and when you run the sample you can input the different options on console to test the different recognition methods.
In this scenario you can directly use the helper class file along with the rest of the sample of conversation transcription to test the scenario.
If an answer is helpful, please click on or upvote which might help other community members reading this thread.