Jump to content

Wikispeech/FAQ

From Meta, a Wikimedia project coordination wiki

What is Wikispeech?

[edit]
  • Wikispeech is an open source text-to-speech solution for the MediaWiki software. MediaWiki is the software used by Wikipedia and thousands of other wikis.
  • We are both combining a number of freely licensed components developed by others, and building a bunch of new stuff.
  • Our initial goal is to launch Wikispeech on the Swedish, English and Arabic Wikipedia language versions. Then we will continue with all the rest of the languages (we will prioritize what language to continue with based on where there are interested volunteers and partner organizations, and/or external funding available).

What is special with Wikispeech?

[edit]

There are a number of reasons why we decided to develop Wikispeech and that make it special:

  1. Our focus is global, and particularly towards less developed parts of the world. Commercial actors have a limited interest in investing in languages spoken in poorer areas. For us, on the other hand, improving accessibility in those languages is something we see as crucial. Wikimedia Sverige's vision is that everyone should have access to the world's collective knowledge. Wikipedia currently exists in 294 different languages and commercial text-to-speech solutions are missing for most of those languages.
  2. Wikispeech is focusing on longer texts. Text-to-speech solutions usually focus on short sentences.
  3. It is possible through crowdsourcing to improve the lexicon so that it sounds better. Users can contribute and make it better, just as they can improve other parts of Wikipedia. Using crowdsourcing is something we think will be very important to quickly increase the quality of speech synthesis in various languages.
  4. It is possible to add more languages. It is built in a modular way to make it easy to scale.
  5. It is built entirely on open source software. This make it possible for us to integrate Wikispeech directly on our servers (as Wikimedia only hosts open source solutions). That Wikispeech will be a server side solution means that people doesn't need to download anything, but that Wikispeech is available directly on Wikipedia. This is important in many countries where Internet cafes and rental mobile phones still are common. That Wikispeech is an open source software also means that other open source projects can reuse the things we develop and hence increasing our impact.
  6. We do not collect data about individual users. We believe that it is none of our business what individuals read about on Wikipedia. (In contrast to large cooperations who has a business model built on the collection of user data).

How did you come up with the idea?

[edit]
  • Staff members at Wikimedia Sverige have worked previously with people with disabilities. We saw the value and importance of developing tools for visually impaired, and proposed the project.
  • The Wikimedia Movement strives to make information accessible, a commitment that has increased with strategic aims in areas such as knowledge equity.
  • Sweden is far ahead in this field and there are a lot of discussions on how to make internet more accessible.
  • We had the opportunity to do a thorough investigative study before we started.

Why is Wikimedia Sverige working on this?

[edit]
  • There is a lot of know-how in Sweden.
  • There is interest and support in the project from the Swedish government.
  • There is a dedicated team working with Wikispeech at Wikimedia Sverige.

Who are you working with?

[edit]
  • STTS - a speech technology company.
  • KTH Royal Institute of Technology in Sweden.
  • We consult Disability organizations in Sweden.
  • We are financed by the Swedish Post and Telecom Authority (PTS).
  • We are also coordinating with staff at Wikimedia Foundation.
  • We are working to engage volunteers in different capacities.
  • There is a huge interest from universities all around the world to join in.