I am interested in using the annotation set feature that is part of model evaluation in the Alexa developer console. Currently, I am able to import in the zip file containing all the mp3 files, however the "Expected Transcription" for each mp3 file is not filled in. According to https://developer.amazon.com/en-US/docs/alexa/asr/create-annotation-set.html#metadata, I can use the bulk edit feature to fill in each file's metadata at once instead of doing it one at a time. However, when I try to import the bulk edit file (I have tried both CSV and JSON) the mp3 file does not get the "Expected Transcription" line filled in. Instead a new annotation is created where the mp3 file is then missing. I have included a picture below for better reference. I have tried uploading the bulk edit first and then the mp3 files but that does not give me better results.
What the annotation set on ASR in developer console looks like: 
