Tutorial: Proofreading your files with text-to-speech

This article will provide you a step-by-step guide on how to “listen” to your translation to proofread it more conveniently.

Proofreading is an integral part of any translation: ideally, it should be done once by the translator (preferably after resting for a few hours), and once by a separate proofreader. However, our eyes often tend to play tricks on us and glance over important parts that even QA tools and spellcheckers can’t catch. The good news is, we also have ears, that are much less likely to miss 🙂

Unfortunately, most CAT tools are not currently capable of reading segments with text-to-speech, so you’ll have to use a workaround and export your document first. Here’s what you need to do:

Note: this method was tested on RTF files exported from MemoQ with Windows 10 text-to-speech language packs. If you happen to know a more efficient method, feel free to chime in!

Step 1: Exporting and cleaning the file

After you’re finished translating, export your document in RTF format by right-clicking it and selecting “Export Bilingual”:

Then, open it in Word. If you’re working on a complicated project, it should look something like this:

That is… less than optimal, and there is no way your text-to-speech will be able to read it back to you in this state. So first off, we are going to remove the Comments column by right-clicking it and selecting “Delete Columns”:

This process might take a few seconds, but you should then be left with only the source and target:

That’s already much better, but as it is often the case with game projects, all the segments are riddled with tags (which have thankfully been simplified during the export process). We are now going to remove all of them with a simple mass-replace. To do so, you have two solutions:

  1. Removing the tags manually
  2. Running a script to delete all the text with a particular color (in this case, the tags)

The first solution is rather straightforward: select any given tag (such as [1} or {2]) and use Search & Replace (CTRL + H) to delete them:

The second solution is if you have a lot of tags and don’t want to waste time removing them one by one, or if you have multiple documents to clean. Open the Macro view by looking it up in the search ribbon (or ALT+F8 by default):

And paste the following code:

Sub FixColor()
    With Selection.Find
        .ClearFormatting
        .Font.Color = RGB(128, 0, 0)
        .Text = ""
        .Forward = True
        .Wrap = wdFindContinue
        .Format = True
        .MatchCase = False
        .MatchWholeWord = False
        .MatchWildcards = False
        .MatchSoundsLike = False
        .MatchAllWordForms = False

        .Replacement.ClearFormatting
        .Font.Color = RGB(128, 0, 0)
        .Replacement.Text = ""
    End With
    Selection.Find.Execute Replace:=wdReplaceAll

End Sub

After saving it, you can now run the macro by pressing the Play button or F5. Once it is completed, you will be left with just the plain text and no tags:

So to recap, all you need to do is:

  • Export the file
  • Remove the Comments column (and any other superfluous columns)
  • Remove the tags

This whole process can look daunting and time-consuming at first, but once you are used to it, it takes less than one minute to complete.

Step 2: Using text-to-speech

Windows 10 comes with a free built-in text-to-speech feature, but if your Windows language is not the same as your target language, you may need to download additional language packs. To do so, simply search for “Text to speech” in the Windows bar, open the Speech menu and scroll down to “Manage voices”

After this, go back to Word, select your target language column (you can also delete all the other columns, but that’s more time-consuming) and select the “Read aloud” option. You can also pick different voices and adjust the reading speed in the right-side menu:

And voilà! You can now close your eyes, take a breather and listen to your translation being read out loud to you 🙂

Granted, I don’t use text-to-speech for UI/system texts, but I find it particularly useful for dialogues and narrative where flow is important. Hope you find it helpful as well!

Comments

  1. Sandrine Guyennet

    Hi Lucile!
    I like using TTS too! I use Natural Reader, which has a few different voices, male and female, and sounds very natural, just like they claim.

    About the tags, I clean the exported file manually in Word before using TTS. I find it faster :
    – I batch search/delete (or replace by a space) tags like
    – I replace all placeholders by real words: [player] becomes Mike, [city] becomes Paris, etc.

    I was told of an open source standalone tool that you could customize to use different voices in dialogues, but I can’t for the life of me remember how it’s called :/

    1. Post
      Author
  2. Marcos Carou

    Hi Lucile!

    I just found your blog yesterday and this article is very interesting. I just wanted to add one comment on the export step: if you click “next” instead of “export”, memoQ offers you several options, like skipping the comments and status columns and deformatting the tags, which simplifies the later process.

    1. Post
      Author
      Lucile Danilov

      Hi Marcos! I didn’t know about the extra step in the export process, which does indeed make it easier. Thanks for the tip!

Leave a Reply

Your email address will not be published. Required fields are marked *