Creating a Source Language Template without a Dialog or Script

Overview

The process examples in this page provide high-level guides when more than one tool is needed or the same tool is used multiple times to perform a task.

While Language Studio’s Automatic Speech Recognition (ASR) is very good, there are many factors that could produce incorrect results in a transcription file. These could range from simple misspellings of a person’s name (i.e., Lindsay vs. Lindsey) to completely wrong words (usually caused by background noise such as explosions, music, etc.) This process utilizes the transcription tools and then the time/dialog alignment tools with a human verification and correction step in the middle.

This example shows the process of creating a subtitle or caption to be used as a Source Language Template when a dialog or director’s script is not available. Typically in this case, the only media file is either a video or audio file.

Prerequisites

• A style guide that would typically be provided to a linguist for the purpose of stylizing a subtitle or caption.

Inputs

• Video/Audio file to be processed.

Outputs

• Subtitle or caption file that is correctly transcribed and formatted to match the style guide.

Steps

1. Setup

If you have already configured the settings in the past then you can skip to step 2.

1. Configure the Transcription Profile
   1. See Configuring Transcription Profiles
   2. Ensure that you enable diarization and set the diarization type to Speaker as distinguishing different speakers will enable the automatic insertion of speaker change markers in subtitles and captions.
   3. Set up any specific runtime glossaries that you want to use to ensure the correct spelling (i.e., Lindsay vs. Lindsey) or for words that the ASR may have difficulty transcribing.
2. Configure the Subtitle/Caption Profile
   1. See Configuring Subtitle/Caption Profiles
   2. Adjust the Subtitle/Caption Profile to match the settings required for the output subtitle or caption.

2. Transcribe the Audio/Video File

This step transcribes the video/audio file into a draft dialog format that is suitable for human review.

1. Submit the video or audio file for processing using the Transcribe Draft Dialog tool.
2. Ensure that you select the correct Transcription Profile.
3. Ensure that you select Draft Dialog as one of the outputs.
   1. It is recommended to output a minimum of Start Time, Speaker and Text in the output columns.
4. There are 2 files in the outputs that will be used in the next 2 steps.
   • <yourfilename>.voice.json – This is the voice data file that will be used to generate the subtitle/caption in step 4. This file contains all the time cues and other relevant data for each word that was spoken.
   • <yourfilename>.dialog.tab – This file is a simple tab delimited file that contains columns representing what was transcribed from the video/audio file.

3. Human Review

This stage provides an opportunity to review the automated transcription and correct any errors, misspellings and mishears.

1. Use a spreadsheet tool such as Google Sheets or Microsoft Excel to review. Simply copy and paste the file into the spreadsheet.
2. Perform a quick scan for anything obvious such as misspelled names and perform a search and replace to correct them throughout the file.
3. Read the text and make any obvious corrections. This can often be done without reference to the original video/audio source.
4. If you need to hear the original audio to verify or fix an issue, use the start time and jump to the time within the video using your preferred player.
5. It may be useful to start from the top of the video and play at 1.5 or 2 x speed while quickly checking the text along with it.
6. Once you are complete, copy the content (be sure to include the column headings) into a text file and save it. 
7. We suggest saving it as <yourfilename>.dialog.edited.tab

4. Generate the Source Language Template Subtitle/Caption File

This stage generates the Source Language Template subtitle/caption. 

1. Submit the Voice JSON file and the human edited dialog file to the Time / Dialog Alignment tool
2. Ensure that you have selected the desired Subtitle/Caption Profile
3. On completion of processing, the subtitle/caption file can be downloaded and is ready to use.
4. If you need to adjust some of the subtitle/caption profile settings, make the profile changes and then repeat step 4. It is not necessary to repeat the other steps.