How to edit manifest when specifying multiple sources for one task with AWS Sagemaker ground truth

Question

I would like to create a task to have one worker perform labeling of multiple sound sources with AWS Sagemaker ground truth. I created a manifest file as follows, but I cannot specify multiple sound sources with source-ref. How to create a manifest file?

dataset.manifest

 {"source-ref":["s3://sagemaker-sample/audio_01.wav", "s3://sagemaker-sample/audio_02.wav"]}

Error

ClientError: Manifest: s3://sagemaker-sample/dataset.manifest has invalid format at line number 0. Make sure that source or source-ref field contains a string value

Answer 1

Specifying multiple sources as a list under a single "source-ref" is not supported. For the actual format, please refer to https://docs.aws.amazon.com/sagemaker/latest/dg/sms-input-data-input-manifest.html. Each line represents a reference to a single S3 file as show below. Example:

{"source-ref": "S3 bucket location 1"}
{"source-ref": "S3 bucket location 2"}
   ...
{"source-ref": "S3 bucket location n"} 

For your case, it would be

{"source-ref": "s3://sagemaker-sample/audio_01.wav"}
{"source-ref": "s3://sagemaker-sample/audio_02.wav"}