Purple DS Text To Speech

Modified on Wed, 05 Aug 2020 at 11:55 AM

General Information


If a PDF issue has been processed using the XML workflow for a readmode, you also have the option to activate an audio read aloud mode. In this way, readers can have the complete issue or individual articles read out to them. 


For this feature, Purple DS uses the latest language algorithms from Amazon Polly



Requirements


The main prerequisites are having an XML-based readmode set up for your publication and an Amazon AWS account.


If you want to use Purple DS TTS (Text To Speech) you also need to decide whether processing should be done via Sprylab's AWS account or your own AWS account.



Processing by Sprlyab

In addition to the one-time setup fees and conversion costs on Amazon's side, Sprylab will charge a service fee for processing



Self processing

If you have your own AWS (Amazon Web Service) account, there are no additional costs apart from the one-time setup fee and the conversion costs on Amazon's side.


In this case we need the following information to test the correct functionality: 


  • Access Key and Secret Key of an API user who has both the rights to use Polly and to perform the necessary operations on S3, for example AmazonPollyFullAccess 
  • Bucket Name, Bucket Region and the prefix (path in the bucket) under which the files are to be stored and which allows public access.

Instructions


The read-aloud mode is created automatically based on the XML data of the readmode. Therefore, only little additional information is required for the setup. 



Select the voices

Depending on the language, the customer has the possibility to choose from several languages. Before processing, you must determine which voice is to be used. 


An overview of the available voice per language can be found here.


Customisation of styling

In addition to the color setting, which should ideally be taken from the readmode styling, there is the possibility to adjust the icons used in the TTS.


These are:


  • TTS symbol: Standard headphones
  • Speaker selection: Standard Avatar Woman/Man
  • Reading speed: Standard speedometer


Optimisation of the read aloud mode



The read aloud mode provides a very good quality with correct pronunciation of most terms. 

However, if you wish to have a different pronunciation for certain words, such as names, this can also be taken into account. 


Please let us know if you want to apply this kind of improvements so that we can coordinate the implementation.


For the implementation we need the terms or names concerned together with the phonetic transcription of the desired pronunciation. The International Phonetic Alphabet (IPA) is used for this purpose.


To be able to process the optimisations, we need them in the form of a text file which you can create with an editor like Brackets or Phase5. For each line in the file, the appropriate phonetic transcription can be assigned to a single term. This means, first and last names have to be entered separately. 


In addition, you should also make sure that the most common spellings and declensions are also listed in the file. 


The structure of a line is as follows: <term>=<phonetic spelling> 



An example: Here the pronunciation of the name Peter Miller is to be improved. The text file contains the following entries: 


Peter=ˈpiːtər

PETER=ˈpiːtər

Peters=ˈpiːtərz

PETERS=ˈpiːtərz

Miller=ˈmɪlər

MILLER=ˈmɪlər

Millers=ˈmɪlərz

MILLERS=ˈmɪlərz




Was this article helpful?

That’s Great!

Thank you for your feedback

Sorry! We couldn't be helpful

Thank you for your feedback

Let us know how can we improve this article!

Select atleast one of the reasons
CAPTCHA verification is required.

Feedback sent

We appreciate your effort and will try to fix the article