Translation and dubbing of the movie at home: Hawking. My short story
A year ago, I translated and voiced the film Robert Zubrin Mars: the Underground. On Habré, the publication was warmly received, and the film itself has been viewed more than 200,000 times and still collects scientific discussions in the comments.
Today I would like to provide details about our latest translation. This is a film about the life of Stephen Hawking. In the post I will talk about the translation process (since this time I did not do it myself), about how best to process the sound recorded at home, and how easy it is to translate the captions in movies. And of course the film itself will show.
To begin with, I’ll give some technical details, information about how we started working on the film and what has changed in the approach to translation and voice acting over the year.
')
Six months ago, in the comments under one of our videos, user Roman S offered to translate this documentary picture. Stephen Finnigan film tells about the life of one of the greatest scientists of our time, Stephen Hawking. We looked, we liked it, it was decided to get down to business. Last year, I myself voiced the film by ear, but now, thanks to a little increased popularity, I managed to attract interested people. One of the students of Kiev Polytechnic, Lera, set to work the next day. I asked her to give a comment and describe her work.
But besides the fact that I acquired a professional translator, over the year some things changed: I became more comfortable with sound processing (see below), stopped voicing on subtitles, and changed my approach to recording.
Subtitle voice acting is good where the timing of these subtitles is well-composed. Otherwise, the text is intermittent, and the intonation can not be properly conveyed. So it is better to recite the text from a file, or having before your eyes the next phrase shown on the screen. A couple of tips for those who want to record a long voice acting at home:
1. Drink warm water. Much water. You can tea from a thermos. This helps to get rid of the "plymkaniya" language and hoarseness in the voice, which invariably appears by the end of the first hour of recording.
2. Do not break the record into files, if there is such a possibility. This makes sense only if you are recording sound in an apartment (for example, like me), and you feel that the overall level of ambient noise has suddenly changed for reasons beyond your control.
3. Do not be afraid to stutter and overwrite. Make long pauses between phrases: cutting pauses and cutting phrases is much easier than creating them artificially, trying to stretch the sound.
This is what my first screen looks like during voice recording.
In my previous major work, many viewers complained that the voice sounded "like a can" - muffled and cropped. Let me remind you what software and hard I use. These are Audio-Technica AT2020 microphone and M-Audio Fast Track external sound card, as well as Adobe's convenient software product - Audition.
Over the year I tried many different sound processing tools, including those that the Habr users advised me to do.
As a result, he stopped at the following sequence of actions.
1. Record sound from a microphone.
2. Deleting clicks (DeClicker tool)
3. Noise Reduction
4. Leveling (L2, RCompressor compressors)
5. (optional) Remove loud "S", "Z", "J" and labial through DeEsser and DeBreath.
Compressors and declicers work very simply, but I would like to talk more about noise. The fact is that to remove noise, the user must capture a specific area of the recording where there is “pure” noise, without a voice. At the same time, many people capture just a clean piece of the record, and as a result they get exactly the “canned” effect that I encountered.
To avoid this, open the spectrogram (default hotkey F9), and look at the frequencies. Usually the noise is at higher frequencies than the voice, so you can select a two-dimensional area (not just a piece of the timeline, but also a part of the recording that grows on the frequency scale) that does not cling to the main bass frequencies of the voice. Just draw for yourself a conditional line, above which the voice levels drop significantly, and highlight the noise in a parallel section. In this way, we will cut off the top noise, but leave our useful sound alone.
It looks like this for me
Everything is very simple here, although it can be cumbersome in terms of the number of actions. We find the moment when the inscription appears on the screen, go to the preceding frame, save it as a PNG image, cut off the excess part and “close” the area with the inscription on the screen with a piece of the frame. For example, ready-made files for replacing the inscription on the logo in the series of videos from NASA JPL, “Mars in a minute” , which I translated a year ago, look like:
In total, much less time was spent on work, with a higher quality of work than last year. The translation was not made by me, it took the author about 16-20 hours, the final editing took place “on the fly”, the sound recording took about two and a half hours. Sound editing is another two hours, then it took about four hours to edit the video and lay the sound on the original track. Oh yes, the 1080p movie render in Sony Vegas took us almost 10 hours. Personally, I was satisfied with the result of the work. Well, I suggest you evaluate it yourself:
Enjoy watching!
PS Special thanks to our editor, Nikita Kolimbet, who corrected the translation and kept kicking me.
PPS The sound was cut out in a short segment of the video - at the same time a new YouTube function was tested, which removes songs that violate copyrights.
PPS In the comments asked for details for material rewards. Why not!
Yandex.money: 41001784052094
WMZ: Z772686954269
WMR: R345291690536
Any Visa / MasterCard: goo.gl/OuGVpc
Today I would like to provide details about our latest translation. This is a film about the life of Stephen Hawking. In the post I will talk about the translation process (since this time I did not do it myself), about how best to process the sound recorded at home, and how easy it is to translate the captions in movies. And of course the film itself will show.
To begin with, I’ll give some technical details, information about how we started working on the film and what has changed in the approach to translation and voice acting over the year.
')
General information
Six months ago, in the comments under one of our videos, user Roman S offered to translate this documentary picture. Stephen Finnigan film tells about the life of one of the greatest scientists of our time, Stephen Hawking. We looked, we liked it, it was decided to get down to business. Last year, I myself voiced the film by ear, but now, thanks to a little increased popularity, I managed to attract interested people. One of the students of Kiev Polytechnic, Lera, set to work the next day. I asked her to give a comment and describe her work.
Although the translation of a written text is easier to do than an oral translation, the translation of a film is quite another. At first, until at least some dexterity appeared and I did not penetrate into the film, the translation went slowly and without much pleasure. Moreover, the guys said not to hurry, so I stretched the pleasure for a couple of weeks. And then somehow I got involved and in a day I took about 15-20 films to translate (I would have had more experience, things would have gone faster, of course).Lera Shukatka, translator
The most difficult thing was to translate something about physics, which I do not fully understand. I had to read a lot to get to the point and translate, but I also had to google a lot or watch the articles on Wikipedia on the principle: the English definition is the correspondence of the article in Russian, hence the Russian definition. Sometimes, she even asked Pasha how something scientifically sounds.
The proofreading of the translation was done at the very end, when everything was ready, because before that there was no point: the work would have stopped altogether if I had only edited it. Moreover, in the end the hand was also more crowded and the appropriate vocabulary spun in my head. Unfortunately, I read it only once, but I’m sure if I’ve read it three more times, I would have changed a lot.
But besides the fact that I acquired a professional translator, over the year some things changed: I became more comfortable with sound processing (see below), stopped voicing on subtitles, and changed my approach to recording.
Subtitle voice acting is good where the timing of these subtitles is well-composed. Otherwise, the text is intermittent, and the intonation can not be properly conveyed. So it is better to recite the text from a file, or having before your eyes the next phrase shown on the screen. A couple of tips for those who want to record a long voice acting at home:
1. Drink warm water. Much water. You can tea from a thermos. This helps to get rid of the "plymkaniya" language and hoarseness in the voice, which invariably appears by the end of the first hour of recording.
2. Do not break the record into files, if there is such a possibility. This makes sense only if you are recording sound in an apartment (for example, like me), and you feel that the overall level of ambient noise has suddenly changed for reasons beyond your control.
3. Do not be afraid to stutter and overwrite. Make long pauses between phrases: cutting pauses and cutting phrases is much easier than creating them artificially, trying to stretch the sound.
This is what my first screen looks like during voice recording.
Sound processing
In my previous major work, many viewers complained that the voice sounded "like a can" - muffled and cropped. Let me remind you what software and hard I use. These are Audio-Technica AT2020 microphone and M-Audio Fast Track external sound card, as well as Adobe's convenient software product - Audition.
Over the year I tried many different sound processing tools, including those that the Habr users advised me to do.
As a result, he stopped at the following sequence of actions.
1. Record sound from a microphone.
2. Deleting clicks (DeClicker tool)
3. Noise Reduction
4. Leveling (L2, RCompressor compressors)
5. (optional) Remove loud "S", "Z", "J" and labial through DeEsser and DeBreath.
Compressors and declicers work very simply, but I would like to talk more about noise. The fact is that to remove noise, the user must capture a specific area of the recording where there is “pure” noise, without a voice. At the same time, many people capture just a clean piece of the record, and as a result they get exactly the “canned” effect that I encountered.
To avoid this, open the spectrogram (default hotkey F9), and look at the frequencies. Usually the noise is at higher frequencies than the voice, so you can select a two-dimensional area (not just a piece of the timeline, but also a part of the recording that grows on the frequency scale) that does not cling to the main bass frequencies of the voice. Just draw for yourself a conditional line, above which the voice levels drop significantly, and highlight the noise in a parallel section. In this way, we will cut off the top noise, but leave our useful sound alone.
It looks like this for me
Translation of captions and titles
Everything is very simple here, although it can be cumbersome in terms of the number of actions. We find the moment when the inscription appears on the screen, go to the preceding frame, save it as a PNG image, cut off the excess part and “close” the area with the inscription on the screen with a piece of the frame. For example, ready-made files for replacing the inscription on the logo in the series of videos from NASA JPL, “Mars in a minute” , which I translated a year ago, look like:
Results
In total, much less time was spent on work, with a higher quality of work than last year. The translation was not made by me, it took the author about 16-20 hours, the final editing took place “on the fly”, the sound recording took about two and a half hours. Sound editing is another two hours, then it took about four hours to edit the video and lay the sound on the original track. Oh yes, the 1080p movie render in Sony Vegas took us almost 10 hours. Personally, I was satisfied with the result of the work. Well, I suggest you evaluate it yourself:
Enjoy watching!
PS Special thanks to our editor, Nikita Kolimbet, who corrected the translation and kept kicking me.
PPS The sound was cut out in a short segment of the video - at the same time a new YouTube function was tested, which removes songs that violate copyrights.
PPS In the comments asked for details for material rewards. Why not!
Yandex.money: 41001784052094
WMZ: Z772686954269
WMR: R345291690536
Any Visa / MasterCard: goo.gl/OuGVpc
Source: https://habr.com/ru/post/245187/
All Articles