Follow

Guide to: Running an Audio Annotation Job

Overview

Appen’s audio annotation tool allows users to annotate audio files with segments and events based on a custom ontology.

Screen_Shot_2020-03-04_at_11.40.36_AM.png

Fig. 1: Audio Annotation Interface for Contributors

Building a Job

Data

  • The audio annotation tool supports the annotation of .wav and .mp3 type files. 
  • Audio files are required to be hosted in a CORS configured bucket. 
  • All data access control features are supported for this tool, including Secure Data Access.

CML

As this is an open beta tool, there is no Graphical Editor support for this tool yet. For access to the tool’s CML gem, please reach out to your Customer Success Manager.

Parameters

Below are the parameters available for the job design. Some are required in the element, while some are optional.

  • {{data-column}} (required)
    • The column from your source data containing the audio URLs to be annotated.
  • {{name}} (required)
    • The results header where the annotations will be stored.
  • {{label}} (optional)
    • The question label the contributors will see.
  • {{validates}} (optional)
    • Defines whether or not the element is required to be answered.
    • Accepts 'required'
    • Defaults to not required if not present.
  • {{review-from}} (optional)
    • This will read in existing annotations on an image.
    • If used, a source column with links to the annotations formatted as we output them is required (example format below).
    • May be used to do peer review or model validation job.

Ontology

Audio annotation jobs use a custom ontology editor that is specific to the audio annotation tool. You can access it by clicking into the Ontology Manager link that appears on the right corner of the job's Design Page.

Screen_Shot_2020-03-05_at_9.26.43_AM.png

Fig. 2: How to find the Ontology Manager via the Design Page

Screen_Shot_2020-03-05_at_7.45.03_AM.png

Fig. 3: The Ontology Manager

Every class in the ontology will have the following attributes:

  • Title
    • This is the class name that will appear in the results and to the contributors working in the job.
  • Instances
    • Any class can be enabled to have instances. This is especially useful if you expect there may be multiple unique instances of one class.
      • Example: If there are multiple people speaking in one audio file, instances can allow annotators to enumerate them and annotate separately.
    • Instances are disabled by default.
  • Description
    • Descriptions can be used to give clear instructions to contributors on what you want to be annotated in the class.
  • Color
    • The color of each class.
    • Unlike other annotation tools, the platform will auto-assign annotation colors to each class. There are 16 pre-selected easy to read and use colors.
    • If you would like to add custom colors, you may upload your ontology manually with a CSV file.
      • A sample CSV file is attached to the bottom of this article.

Screen_Shot_2020-03-05_at_7.55.04_AM.png

Fig. 4: Editing a Class in the Audio Annotation Ontology Manager

Reviewing Results

Results

Results will be provided as a secure link to a JSON file describing the annotations.

Please note: Due to security reasons, JSON result links will expire 15 days after generation. To receive non-expired result links, please re-generate the result report. 

The objects in the JSON include:

  • layerID and ID
    • The universally unique identifier of every annotation, along with the ID of the layer it belongs to. All unique instances of the layer will have its own ID so they can be connected across multiple segments.
  • startTime and endTime
    • This will be displayed in seconds, to the millisecond.
  • ontologyName
    • This is the class name the contributor assigned to the segment.
    • If instances are enabled, integers will be appended to distinguish between different instances.
  • annotatedBy
    • This will be "Human" if annotated by a contributor and "Machine" if left unannotated in the judgment by the contributor.
  • nothingToAnnotate
    • This will be Boolean true or false.
    • This will be true if the contributor does not have a judgment they could make.
      • Example: The audio does not include any speech.

Below is an example of the JSON results:

{

"annotation": [

[

{

"id": "10b784ae-74a4-49bd-ba39-c11d086fd7f6",

"layerId": "a7a09d79-7526-43be-a9ae-a3cf1d7a94d9",

"startTime": 111.375,

"endTime": 149.486,

"ontologyName": "Class 1"

“annotatedBy”: “Human”

}

"nothingToAnnotate": false

"ableToAnnotate": true

}

Reviewing Results

To review the results of your job in the Appen UI, you can do the following:

  1. Go to the Data Page of the job.
  2. Click on a unit ID to view the unit page.
  3. A custom widget loaded with the contributor's annotation to review will be displayed.

Screen_Shot_2020-03-05_at_8.00.52_AM.png

Fig. 5: Reviewing Annotations via Unit Page

Additional Notes:

This product is in BETA, so please consider the following important notes:

  1. The job must be set-up in the Code Editor; the tool is not supported in the Graphical Editor yet.
  2. Audio Annotation jobs do not support test questions or aggregation at this stage.
  3. Launching this type of job requires one of our trusted contributor channels. Please reach out to your Customer Success Manager to set this up.

Was this article helpful?
1 out of 1 found this helpful


Have more questions? Submit a request
Powered by Zendesk