Skip to content
Gallery
gull-aa66bde11f73425aacb7d5f6bc5
Aviary User Guide
Home
Getting Started
Aviary Features
OHMS in Aviary
Development Log
Search
Share
Explore
Aviary Features

Transcription Integration

Last edited 602 days ago by Kevin Glick
Automated transcription integration with external service providers like IBM Watson or Trint involve a two step process for creating new transcripts after system configuration:
An organizational user requests a new transcript be created
An organization owner approves the new transcript request, thereby approving the cost involved.

IBM Watson

This integration helps to generate transcriptions against given video and audio.
Aviary allows organization administrators to request automated transcriptions using IBM Watson.
The option to request a transcription is available on the Resource Detail Page view under the 'Transcription' tab. There are three stacked dots that, when clicked, provide the option to "Request new transcript."
Requirements:
The audio/video file should be hosted by Aviary. It is not possible to send embedded content from Aviary to IBM Watson for transcription, e.g., media files embedded from Youtube, Avalon, Vimeo, or Direct URLs.
Before populating the new transcription request module, automated tasks needs to fetch the IBM Watson supported languages list. If the languages list is not available from IBM Watson the organization admin cannot request the transcript.
Options:
An organization user can request a new transcript using IBM Watson. The following options are available:
Language
Speaker Diarization This option will turn on speaker identification.
Smart Formatting This option will generate special formatting for dates, times, series of digits and numbers, phone numbers, currency values, internet email, and web addresses.
Profanity Filter This option will censor profanity from the results.
Remove Hesitation Markers This option will remove "%HESITATION" from the results.
Cost:
Before creating a transcript job, the user can see the cost for the job. The fee will be charged from a connected 'Stripe' account (i.e., stored credit card) for the organization.
Approval:
Once the transcript request is created, it is available in the "Automated Metadata" tab in the Aviary admin sidebar. An organization admin user must then approve or reject the job.
Once the transcript request is approved by an organization admin user, Aviary runs a cron job to check new requests and to run the tasks in the table below. These tasks send the transcript request along with selected options and keep pooling for its status.
Once the job status is Complete on IBM Watson, the tasks below will download the transcription and upload it to Aviary and store it as a child of the parent media file it was requested for; it will also update the status for the job on the "Automated Metadata" tab.
Technical
User facing files:
transcriptions_controller
responsible for managing the organization automated meta process initiation
Worker Processing:
rake aviary:watson:check_job
Trigger the job to check the job status of watson and do the related process
rake aviary:watson:languages
Get the available watson languages and store to the database

Trint

This integration helps to generate transcriptions against given video and audio.
⁠
image.png
⁠
⁠
Aviary allows organization administrators to request automated transcriptions using Trint.
The option to request a transcription is available on the Resource Detail Page view under the 'Transcription' tab. There are three stacked dots that, when clicked, provide the option to "Request new transcript."
Configuration:
Before using the Trint request transcript feature, an organization admin needs to create an account on
https://trint.com/
⁠
and get an API key from Trint and a callback URL from Aviary.
The API key can be added into Aviary: Integrations → Trint (there is more information on the Aviary Trint integration settings page.)
The callback URL will be generated for the user by Aviary and that needs to be given to Trint in the Trint API settings on
trint.com
⁠
for the user's Trint account.
Requirements:
The Trint integration is enabled by adding an API key from Trint and creating a callback URL.
The audio/video file should be hosted by Aviary. It is not possible to send embedded content from Aviary to IBM Watson for transcription, e.g., media files embedded from Youtube, Avalon, Vimeo, or Direct URLs.
Options:
An organization user can request a new transcript using Trint. The following options are available:
Language
Speaker Diarization This option will turn on speaker identification.
Cost:
A user will manage their own Trint plans and pricing on their own Trint account.
Approval:
Once the transcript request is created, it is available in the "Automated Metadata" tab in the Aviary admin sidebar. An organization admin user must then approve or reject the job.
To approve a request, click on the green Approve action button on the right side of the appropriate row for the request.
If you need further information to make your decision, you can click on the blue View button to go to the requested resource.
If you do not approve of the request, you can click on the Delete button.
⁠
image.png
⁠
⁠
It is possible to approve multiple requests for Trint transcript creation, by selecting the box in each row, clicking on the Table Objections button, selecting the Bulk Approve option from the Bulk Edit pop-up window, and clicking on the Apply button. ​
image.png
⁠
⁠
Trint Processing:
Once approved, Aviary will send the job to Trint and once the job status is complete or fails on the Trint side, Aviary will get the information and update the database accordingly.
Trint sends back to Aviary an initial draft transcript which Aviary sets as private, under the assumption that the Aviary organization may wish to review and revise it. An initial, or not fully completed transcript in Trint may have statuses of Ready, Not Edited, or Edited. ​
image.png
⁠
⁠
To revise the transcript, the user must work in Trint, making as many edits as necessary in each paragraph and marking the checkbox to the right of the paragraph to indicate that the revision has been completed. ​
image.png
⁠
⁠
When all the checkboxes have green Xs, then Trint marks the transcript as Corrected and sends a revised version of the transcript to Aviary. ​
image.png
⁠
⁠
The Aviary organization must still change the transcript to public if you wants external users to see it.
If a user wants to make additional edits to the transcript after all the boxes have been checked and it is marked in Trint as Corrected, then the user must go back to Trint and uncheck a couple of boxes, make additional changes, recheck all the boxes, save, and Trint will submit a new version to Aviary.
An organization must maintain the Trint and Aviary account throughout this process to allow the integration to continue to pass data back to Aviary.
⁠
Once the transcript request is approved by an organization admin user, Aviary runs a cron job to check new requests and to run the tasks in the table below.
Task
User facing files:
transcriptions_controller
responsible for managing the organization automated meta process initiation
Worker Processing:
rake aviary:transcoding:check_job
Check the job of transcoding and update information in the system
rake aviary:transcoding:start
Trigger the job for transcoding of the resource media file with status pending

Want to print your doc?
This is not the way.
Try clicking the ⋯ next to your doc name or using a keyboard shortcut (
CtrlP
) instead.