Video tutorial on how to perform the QA stage and provide corrections and comments when submitting the task:
You will be sent:
1. link to Google Drive to the edited video of the film you worked with
2. comments from the editing team about the corrections they ask you to make
3. link to your project on the platform (so that you could make corrections there)
Please review the already edited translated video for errors and inaccuracies made by the editing team and any issues or errors that you might have missed in the previous stages of working with the translation and audio tracks. Please take note of all the errors in the translation (and correct them) and everything you would like the editing team to change.
Errors in audio tracks (translation/pronunciation)
Errors in audio tracks can include:
Grammatical errors;
Stylistic errors;
Wrong language (e.g., not the Spanish "no" but the English "no");
Incorrect pronunciation;
Incorrect gender (e.g., addressing a woman as if she were a man);
Incorrect speaker voice (e.g., a woman is speaking on the screen, but the voice is male, and it's not just a voice-over translation of an unknown language);
Different pronunciation of the same name/surname in the film;
etc.
To correct the translation, please regenerate the necessary tracks in the project, mark them with checkmarks in the project (see the video), and also write the segment numbers to the manager when submitting the task. If for some reason you do not have access to the project on the platform, request access to it (link).
Please note that the sudden change of voice of one character (see the video below, doctor is speaking in 3 different voices) should also be resolved, as this is considered to be a mistake. This can be changed by regeneration of the track on the platform or even changing the voice ID of this emotion of the character to the more suitable one.
Errors made by the editing team
You may also find errors made by the editing team:
The English text is heard alongside the translated text (the English was not muted in this moment in the video);
The line starts too late/too early (e.g., the character's mouth is moving for a few seconds, but there's no sound);
A segment's translation is missing (the character's mouth is moving, but there's no sound);
The original film's sound was not inserted (e.g., laughter that the neural network could not generate, and you see the character laughing on the screen, the English video has the sound of laughter, but the translated video does not);
etc.
To provide comments to the editing team, you need to write the timecode and the error, as well as the necessary action the editing team needs to take (see the video).
Example: 00:02:03 – the line starts too late, it should start a couple of seconds earlier.
What if the translation of a line is missing, and it was not originally in the project?
If the translation of one or more segments is missing, you need to watch this segment in the original English version of the film, write down the timecode, the translation, and generate this track with the appropriate Voice ID.
There are two ways to do this.
Option 1 (when there are few such tracks):
Take the necessary Voice ID from your project (you can find it by clicking on the regeneration button; note: nothing will regenerate immediately, you will just copy the needed Voice ID and exit this window).
Next, select the function Synthesis by Voice ID
Enter the necessary Voice ID, the translated text, and click SYNTHESIZE
Achieve the desired pronunciation and approximate length of the phrase. You can see what the additional parameters do by clicking on the question mark icon next to them. They are set to default values, but you can change them if needed.
When you have achieved the desired sound and length of the audio, download the resulting track and rename it in the format 1_character name_timecode in the format 00.00.00.000. Example: 1_Maggie_00.40.57.000. (00 hours, 40 minutes, 57 seconds, 000 milliseconds)
Please don't forget to send and describe such missed and voiced lines when submitting corrections and comments to the manager.
Option 2 (when there are many such tracks):
Create an Excel spreadsheet with the translated missing lines. A sample is shown below.
Let's go over what each column represents.
A. Timecode 1 - the opening timecode (the time when the line starts)
B. Timecode 2 - the closing timecode (the time when the line ends)
C. Block - this is the segment's number
D. Speaker - this is the character's name
E. Original Text - this is the original line (in our case, in English)
F. Machine Translation - you don't need this column, leave it empty
G. Translation - this is the translation of the line that you enter
To simplify the work on the missing lines, the following algorythm is suggested:
Fill in the correct opening timecode in column A (without milliseconds, just hours:minutes:seconds,000).
In column B, enter the timecodes that will be the opening timecodes of the next line (they are not important, but the system requires them to be there).
In column C, start the numbering from 1 and drag it down, i.e., it will be 1, 2, 3, and so on.
In column D, write the speaker's name according to the name of the character in your project. That is, if the character's name is Tom, write Tom in the row with his line.
In column E, I suggest you enter the original line for simplicity. The system requires the original text to be there, otherwise, the Projects button won't work.
Leave column F empty.
In column G, enter your translation of the missing lines.
Here is the template you can fill out: Table draft.xlsx
Remember, the timecodes in the table need to be written as hours:minutes:seconds,000
Example: 00:40:57,000 (00 hours, 40 minutes, 57 seconds)
If you encounter formatting issues with the timecodes in the table, I recommend setting the cell format to Text.
The table is ready, you have filled it out, and you want to check if you have filled in all the columns correctly. Use the Check XLSX tool.
Now you can proceed and create the project.
Go to the Projects tab
Select Create a new project from XLSX in the upper right corner of the page
In the project name, specify the language abbreviation and the name of your film, for example, ES_Trixie
Attach the file
Do not use the Project Translation slider
Click on the UPLOAD button
On the left side of the page, find the Synthesis section, and click on the WITH VOICE ID button
In the Speakers section on the left, enter the necessary Voice IDs of the characters (see the 1st point of the scheme in the "Option 1" for the missing tracks)
In the Synthesis section, click on the Run button
After the audio generation is complete, click on the Edit button in the upper right corner of the project page
Then, listen to the resulting blocks, edit, and regenerate them, if necessary, until the desired result is achieved
Don't forget to send the project link and describe the missing and voiced lines when submitting corrections and comments to the manager. You can copy the project link from the browser's address bar.
1 more ways to voice the missing lines:
If you can hear what is said (here we work with the subtitles, but you can just take the timecode from the video)