Fix text markup for Chinese audio data.
Budget: $30 – $250 USD
In the task, you need to correct the incorrect annotation.
The archive contains two folders (“wav” and “docx”).
The “wav” folder contains audio files; folder "docx" - textual information of what is said on the audio.
As for fixing the .docx files themselves:
1. There may be “unintelligible”, “inaudible” and other segments with similar names segments in the text. In case you cannot recognize them, highlight them in gray.
2. If you can recognize a segment marked as “unintelligible” or “inaudible”, then you need to highlight it in red and write the recognized text, highlighting it in green.
3. If you find that the text does not match what is said on the audio, then the design rules are the same as in paragraph 2.
4. It is also necessary to correct the spelling of numbers to their textual form in the same form in which it is pronounced on audio. Correction should be underlined with colors. Example: 5 five.
5. If email, website, etc. are written in the text, then they must be written as they are pronounced. Example: if on a record a person pronounces the mail "my-mail @ gmail.com" as "my hyphen mail at gmail dot com", then it is with this form that the name of the mail should be replaced. If a person, for example, does not pronounce the word "dot", then, of course, it is not necessary to write it.
6. Words in other languages should be in italics.
At the end of the file, you need to write "Clean version". In it, you need to write the version of the text obtained after all your corrections + segments that you have left greyed out.
The archive contains two folders (“wav” and “docx”).
The “wav” folder contains audio files; folder "docx" - textual information of what is said on the audio.
As for fixing the .docx files themselves:
1. There may be “unintelligible”, “inaudible” and other segments with similar names segments in the text. In case you cannot recognize them, highlight them in gray.
2. If you can recognize a segment marked as “unintelligible” or “inaudible”, then you need to highlight it in red and write the recognized text, highlighting it in green.
3. If you find that the text does not match what is said on the audio, then the design rules are the same as in paragraph 2.
4. It is also necessary to correct the spelling of numbers to their textual form in the same form in which it is pronounced on audio. Correction should be underlined with colors. Example: 5 five.
5. If email, website, etc. are written in the text, then they must be written as they are pronounced. Example: if on a record a person pronounces the mail "my-mail @ gmail.com" as "my hyphen mail at gmail dot com", then it is with this form that the name of the mail should be replaced. If a person, for example, does not pronounce the word "dot", then, of course, it is not necessary to write it.
6. Words in other languages should be in italics.
At the end of the file, you need to write "Clean version". In it, you need to write the version of the text obtained after all your corrections + segments that you have left greyed out.