Automating language learning listening material creation with Google Translate text-to-speech: The technology

  1. A digital audio lab heavily depends on the availability of, but does not usually come with digital learning materials (and recent exceptions are exceptions for a reason)  Some digital audio materials that come with your textbook may be adaptable. “Rolling your own” has all kinds of advantages (allows for personalization, for both teachers to express themselves, and for students to learn), but can be a chore.
  2. Can the LRC find a workaround?  Here is one attempt: making Google translate (too often abused by students in its original interface) text-to-speech (unusable for learning material in its original interface since severely crippled) usable for digital audio learning material production, provided you have a source text in the target language. image
  3. GoogleTTS can serve as the gateway to better suiting Google Translate text-to-speech features to the needs of the LRC:
    1. imageGoogleTTS allows for arbitrary-length input text (it chunks it automatically).
    2. GoogleTTS produces intermediate local audio files which we can postprocess.
    3. Google Translate’s automatic language recognition remains a sore point: it is not reliable. Unlike Google Translate, GoogleTTS has no interface to set the language manually when the automatic recognition fails.
  4. Batch-download the files from Google Translate, using MS-PowerShell: <
    $global:folder = 'G:\Temporary Internet Files\Content.IE5'
    $filter = '*.mp3' # &lt;-- set this according to your requirements
    $global:destination = 'G:\conf\programs\GoogleTTS\mp3'
    $global:path
    $global:path1
    $currenttimeFunction MonitorAndMoveFile{
    $fsw = New-Object IO.FileSystemWatcher $folder, $filter -Property @{
    IncludeSubdirectories = $true # ja, brauch ich für googletts i&lt;-- set this according to your requirements
    NotifyFilter = [IO.NotifyFilters]'FileName, LastWrite'
    }
    $onCreated = Register-ObjectEvent $fsw Created -SourceIdentifier FileCreated -Action { # the even monitored is file created - to force recreation of files by googletts, you may have to clear watched folder of all mp3 &lt; 100kb first
    $global:path = $Event.SourceEventArgs.FullPath
    Write-Host $global:path -ForegroundColor Magenta # this works also
    $name = $Event.SourceEventArgs.Name
    $changeType = $Event.SourceEventArgs.ChangeType
    start-sleep -Seconds 2 # The OnCreated event is raised as soon as a file is created.
    if ($global:path -ne $global:path1) # it is a createdevent on a different file from last time - just in caseon oncreated not firing clear cut, but it seems to
    {
    $currenttime = Get-Date -Format yyyy-MM-dd-hhmmss
    Write-Host "attempt copy $global:path1 to $cuurrenttime" # try copying the past file
    # Copy-Item -Path $global:path1 -Destination "G:\conf\programs\GoogleTTS\mp3\$currenttime.mp3" -Force # that worked with the last generated file, wait: the last one is the one that remaisn behind, earlier ones get overwritten
    Copy-Item -LiteralPath $global:path1 -Destination "G:\conf\programs\GoogleTTS\mp3\$currenttime.mp3" -Force # that worked with the last generated file, wait: the last one is the one that remaisn behind, earlier ones get overwritten
    # use parameter -literalPath because files in the temp folder have usually [ and ] inside the name which acts as wildcards characters
    $global:path1 = $global:path
    }}
    while (1) {
    sleep -Milliseconds 100
    write-host $global:path # this works
    }}
    MonitorAndMoveFile
    #Unregister-Event -SourceIdentifier FileCreated
    
    
  5. Merge the downloaded files (wisely numbered sequentially):
  6. image
  7. Fix minor errors in your audio editor:
  8. image
  9. Done:
    1. Here I have a lot of questions for a speaking exam in ESL, and with a much better accent than my own.
    2. Nifty, plus output sounds even better for German than for English. Note, there is no attempt to parse sentences semantically. Some languages chunk better than others (I made some little improvements in this regard to the original program). Other common problems include numbers and in German I find myself, when listening, tending to look up once in a while and shake a high school students by the shoulders, asking him: “Do you actually understand what you are reading?!” Smile– which in my eyes is an indicator to the progress made in speech-synthesis.
    3. Other examples include French,
    4. Hindi,
    5. Italian,
    6. Spanish.
  10. So can the LRC relieve teachers from recording their cue files for the digital audio lab listening comprehension and exam? Within limitiations.
  1. No comments yet.
  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: