AI Processing

When you register a podcast in LISTEN or start a podcast in LISTEN, it is automatically processed by the AI. These processes may take several minutes to several hours after processing begins. Also, depending on the details of the process, the time until your turn to be processed may vary.

This page describes the processing of AI in LISTEN and how to correct it.

Transcription

When a new podcast is submitted, it is automatically transcribed. Transcription is basically done in the order of the episode's pub_date. Therefore, the latest episodes are processed relatively quickly, while older episodes may take several days or weeks to be transcribed. (Options to speed up the processing of older episodes are being considered.)

Editing transcribed text

Select the pencil mark in the upper right corner of the text you wish to edit.

Edit the selected text. (When editing is complete, press enter key to confirm.)

Chapter Separation, Table of Contents and Summary Creation

Once the transcription is complete, chapter separation, table of contents and summary creation will follow. 

Chapter separation, table of contents and summary creation will only be performed for the latest episode of a certain length (An option to create chapters and a table of contents for all episodes is under consideration.).

Editing chapter-by-chapter headings

The results of the chaptering and headings, as well as the text, can be edited using the pencil symbol in the upper right corner.

You can also move the position of headings by editing the timestamp.

Speaker Separation

If two or more performers are added to an episode that has already been transcribed, speaker separation is automatically performed.

Registering and Editing Performers

Access the episode page while logged in with the podcast administrator account, and select a performer from the "+" mark.

Edit the results of speaker separation

There are two stages in the process of linking and editing text, speakers, and performers.

(1) Linking text, speakers, and performers

After automatic processing by the AI, the transcribed text is linked to each speaker. The assignment of speakers and performers is also done automatically.

To change the performers assigned to each speaker, click on the downward pointing mark next to the performer's name and select from the performers displayed below.

(2) Correcting the ties between the text and the performers

Since transcription and speaker separation are performed automatically by AI, some parts may be incorrect. 
To correct details, select the text you wish to correct and select the pencil mark in the upper right corner of the screen.

Select the downward pointing mark next to the performer's name.

From the list of performers, select the performer you wish to show.

Then, you can correct the connection between the text and the performers only in one part.