9 Tips for Localizing Audio

Captain C-3PO by Jeff Nickel

If you dealt with audio recording - whether it’s a character voice for a game or a narration for a video, you’ve probably noticed that it’s not cheap. It is important to do everything right the first time to reduce costs. The same goes for audio localization: each error is multiplied by the number of languages. In this article, we will share tips on how to interact with recording studios and localization services, how to optimize and speed up the process, reduce the risks, and at the same time, the costs of sound localization. It does not matter if you order these services from us at Alconost or another company - knowledge of all the pitfalls will definitely come in handy.

1. Design Matters

Proper script design ensures that the recording runs smoothly and without problems. An ideal variant is a script in the form of a table , where each line corresponds to a single sound file (here is a vivid example of a script for a game video ).

At Alconost, we are sometimes faced with situations where a client only wants to record a narrator in a foreign language, leaving all the work to bring the video and audio sequence along. In this case, the customer has to work with audio recordings in a language he does not know, and we, for our part, are doing everything we can to simplify this task for the client:
')
- when transcribing, we ask the editor (carrier of the source language) to mark the timing (which phrase is spoken at what second) and break the text into phrases (for example, a new scene - a new table cell). Transcribing itself is the process of “removing” the text from the original entry, i.e. transformation of the text pronounced into printed text;

- when translating, we ask the translator (carrier of the target language) to keep the original breakdown (to be clear: here is a fragment in the original - and here is a translation of this particular fragment);

- when recording, we ask the announcer to break the record into scenes (for example, replicas for each new scene are saved in a separate file).

2. Create a list of characters and describe their character.

If you need several voices, create a list of characters and their short description, which includes name, age, gender, and other characteristics. This will help the recording studio to provide an adequate selection of speakers for these roles, to ensure the necessary number of them in different languages, and to avoid such mistakes as recording a female voice instead of a male one or vice versa. By the way, the gender of the character does not have to coincide with the gender of the announcer: professional voice actors with a wide creative range can successfully voice both a tomboy-boy and a respectable lady.

The character description of the character is important so that the announcer imagines him better and can play well. For example:

Character A: 40 years old man, good-natured, rustic (farmer), “your boyfriend”. Friendly with his (but without joking), with strangers with strangers (but not ham). The voice is powerful, but not harsh. The rate of speech is normal.

Or:

Character-raccoon: a tall voice, like a child, but laughs and releases snide remarks with hoarseness. Speech rate is very fast, but clarity can not be neglected. You need five variants of laughter (kind, malevolent, hysterical, demonic, loud laughter) and three options of exclamation from fright.

Ideally, a screenshot or video capture is included in the task, where the character is visible.


^{Melancolia by Kristina Alexanderson}
If there is an opportunity to give the announcer to listen to the recordings of the same characters in any other languages, this is always a plus. Clarifications are welcome: “not as fast as in the reference” , “exactly as in the example” .

What if you neglect the descriptions of the characters? In an incomprehensible situation, the announcers will either ask you questions (this is best, but the exchange of questions and answers may delay the deadlines), or go to their own discretion (and will be right in their own way). In the second case, if you do not like the speaker’s version, it turns out that the formal criteria are met, and your claim concerns a point that you have not specified. Then dubbing may not be free, although announcers often try to meet the customer.

3. Limit the number of votes

Often people want to spend on localizing audio recordings less money than on original audio. As announcers usually have high minimum bids, limiting the number of votes is a good way to cut costs. For example, a video showing 12 different interviewees may well be voiced by two professional speakers: a man and a woman.

4. Avoid audio files with multiple characters at once.

In a perfect design, the audio recording process is quick and easy. The assignment is set in detail and accessible, the speakers are recorded, the recordings are checked by native speaker linguists and cleaned of unnecessary sounds and noises (such as page turning); the replicas are cut into files, and the files are named according to the terms of reference.

When you have one audio file that contains several announcers at once, everything becomes much more complicated. Since speakers are always recorded separately, audio files are split and recombined in the final stages of the process. This increases the risk that something will go wrong. After all, the engineers who carry out these manipulations are unlikely to speak all the foreign languages ​​in which records are made.

5. Prepare a pronunciation guide.

If you read the text out loud, you will soon realize that some words may be spoken differently. Especially abbreviations and abbreviations. Foreign languages ​​make the task even more difficult.

Some pronunciation rules are common, but some are defined by the company itself. Your studio should study the script and make a list of all the words in which pronunciation can be different . Translators and editors should clarify all points at issue before recording.

If there is a time limit on any cue - Captain Obvious suggests that it is desirable to warn the announcer before recording.

6. Leave air in your video

If you are familiar with localization, you probably know that English is one of the most compact languages. If we translate the same text into Russian, it will increase by 10 percent, and if into French or Spanish, even by all 20 (we even dedicated a separate post to this with the statistics we collected).

If you do not have the opportunity to change the duration of the video sequence, then it will be difficult for you to fit the localized sound under the video without speeding up the speech or shortening the original text. Both of these options can degrade the perception of your video. Moreover, the text or action in the frame may not coincide in meaning with the localized audio. This is especially problematic for videos whose lengthening or editing costs a lot of money.

Therefore, the best option would be to leave some “air” in your original video: add pauses in a few seconds where it can be done. Subsequently, this will simplify and speed up the localization process. For our clients, we in Alconost try to adjust the number of words to the desired length of phrases when translating text or editing it. So it turns out to avoid unnecessary pauses or inappropriate haste in the reading.

7. Stock up on source

You are most likely to be asked for them. But just in case, we recall: you cannot do without samples of the original audio, if you want a voice acting in the style and tone of the original.

If we are talking about video, make sure that you have the source of the video: so you can slow down or speed up the scene a little while localizing, so that the animation gets into the pace of the announcer's speech in a new language. It is desirable at the same time that in the audio track the voice of the announcer, music and sound effects go separately from each other: this way you can slightly move the sound effects if you move the animation.

And one more nuance. When it comes to video, often it is necessary to localize not only the text of the announcer, but also the captions in the frame. If you want to translate them too, it is easiest to do this with the source of the video.

8. Make sure the script and video sequence exactly match.

This is especially important when it comes to videos. The voice acting usually corresponds to the written script, but last-minute edits in commercials often occur. This results in discrepancies between the scale and the visual part. Check out the final video carefully.

9. Choose specialists who know foreign languages ​​and have experience in audio localization.

It happens that the translation of the text for the audio recording was done by your friends, you yourself or the translators who are not native speakers of the target language. All this is an occasion to think about doing pre- reading (reading) performed by a native speaker before reading. After all, some speakers may not even take up the recital, if they see that the text in their native language is written illiterately. Announcers can offer their services for proofreading and this is better than nothing, but ideally it is still worth engaging an editor (carrier of the target language) who subtracts text that is not translated by the carrier. If the quality of the translation leaves much to be desired, or if the meaning in the translated text has changed or been lost, then instead of proofreading, it is better to do the editing (when editing, the translated text is compared with the original).

Working with foreign languages ​​adds complexity to the entire audio recording process. Localization of audio requires utmost attention to the selection of speakers and their compatibility with each other, the design of the script, recording methods, linguistic testing of the result. For example, if you assign localization into Asian languages ​​of a studio that translates mainly into English, you risk finding a recording of a Korean announcer instead of a Chinese one, cut off half-words or other inconsistencies caused by ignorance of the language.

When choosing a service for localizing audio, be guided by the presence in the staff of professional speakers, native speakers and editors, long experience of successful work in this market, and reviews of satisfied customers. We at Alconost are ready to help you with the voice acting and its high-quality localization - even if we are talking about rare languages.


about the author

Alconost is engaged in the localization of applications, games and websites in 60 languages. Language translators, linguistic testing, cloud platform with API, continuous localization, 24/7 project managers, any formats of string resources.

We also make advertising and training videos - for websites selling, image, advertising, training, teasers, expliners, trailers for Google Play and the App Store.

Read more: https://alconost.com