Audio files may include standard metadata, such as agent identification and recorded date. Audio files may also include custom metadata; usually, this metadata is defined by the customer using the recording system.

The following is a general workflow for using custom metadata when processing audio files:

1. Establish metadata requirements for your processing flow. Take note of how special fields (agentid, datetime) are populated during processing and whether any additional custom fields will be required for your use case.
2. Configure a folder's custom metadata fields individually using the UI or in a batch by uploading a file.
3. Prepare the metadata files in JSON or XML format, noting metadata restrictions and formatting and syntax requirements.
4. Upload metadata files either individually or with audio using the V‑Spark UI or with the /transcribe endpoint.
5. Search or view the custom metadata using the UI or in transcription output.