- From: Janina Sajka <janina@rednote.net>
- Date: Wed, 5 May 2021 08:33:15 -0400
- To: public-pronunciation@w3.org
The following announcement may be of interest to this group. Also, Willem may be a developer we might want to invite into our proceedings. Willem van der Walt writes: > Good day, > Please feel free to redistribute to individuals or organizations you think > might be interested. > My appologies if this is somewhat off topic for this list. > Kind regards, Willem > > ???Announcing the ebook augmentation system > ======================================== > > Introduction > ------------ > The Voice Computing Research Group at the Council for Scientific and Industrial Research in South Africa has developed a system > which automates the addition of either human-narrated or synthesized speech to a standard EPUB 3 publication. > The system automates the alignment of the speech to the text of the book at paragraph, sentence and word level, allowing the > eventual user to switch among these levels of granularity on the fly. > > In short, the user of the system pushes in the standard EPUB 3 and, optionally, the pre-existing human-narrated audio files to > get out an EPUB 3 with the synchronized audio included. > When no human-narrated audio is available, the text of the book is synthesized, and the synthesized audio is added instead. > The system also allows for a combination of human-narrated and synthesized audio. > It is operated through a user-friendly web-based interface. > > Multilingual support > -------------------- > As the definition of the main language of standard EPUB 3 files is often incorrectly specified in practice, > the user selects the main language before the processing starts. When passages in one or more foreign languages exist in the > book, these must be marked up in the input EPUB 3 file with the xml:lang tag. > This is required to ensure proper alignment of the audio with the text and, in the case where speech has to be synthesized, > to enable the automatic selection of a TTS voice in the correct language. > > By default, the Qfrency Text-to-Speech (TTS) synthesizer (also developed by the same research group) is used, with fallback > to the open-source Espeak synthesizer when a language not supported by Qfrency TTS is encountered. > With some customization, it is possible to use other TTS engines as well. As a working example of this, the open-source RHVoice > TTS engine was implemented. > > Input and output formats > ------------------------ > Once a book is augmented with audio, the following output formats are available for download: > 1. An EPUB 3 file with the audio added and synchronized to the text. > 2. In the case of synthesized audio, a ZIP file containing a set of MP3 files with just the audio. > 3. A ZIP file containing a portable embosser format (PEF) file for Braille production. > This is obtained by running the DAISY Pipeline2 product in the background. > 4. A ZIP file, also produced by the Pipeline2 product, with a DAISY 2.02 version of the book. > At the time of writing, the latter still has some issues which will hopefully be resolved soon. > > As a convenience to the user, a simple web-based interface is provided for conversion of other input formats into EPUB 3. > Currently, DOCX and PDF can be converted. > > Reading the resulting EPUB 3 books > ---------------------------------- > The books can be read using any EPUB 3 reader that supports media overlays. With the exception of Readium, most of the ones > available, however, do not support the on-the-fly changing of granularity. We have an EPUB 3 reader which is in beta. > It supports the multi-level granularity (paragraph, sentence and word) and has some additional features like search, repeat > etc. It currently runs on Android and Windows, with iOS in the pipeline. > > Invitation to pilot the system > ------------------------------ > We want to extend an invitation to interested providers of accessible reading material and publishers world-wide, to contact us > to participate in piloting the system. > > Each participating organization will receive one or more accounts on the system through which it will be able to upload its > books for augmentation securely. The books uploaded and processed by each account are only visible to that account holder. > During the pilot phase, the system will run on our servers with the aforementioned TTS engines as options. Other implementation > options will be available when the system goes into production. Our EPUB 3 reader will also be provided to the participants. > > To participate in the pilot, please email a request to: Ilana Wilken <iwilken@csir.co.za> with the subject: > "Ebook augmentation system: international pilot". In the body of the email, please provide the names and email addresses of > the individuals in the organization who will require accounts. Optionally, indicate the language(s) which you would > like to process with the system. For technical enquiries about the system itself, please send an email to: > Willem van der Walt <wvdwalt@csir.co.za>. > > We would like to achieve the following objectives through the pilot: > 1. More real-world books through the system, with feedback from real-world users on both the system and on the resulting > output books. > 2. Suggestions on the prefered business model, e.g. a once-off license or annual subscription license, maintenance and support, > etc. > 3. Whether you would prefer to run the system externally over the internet through a web interface (like in the pilot) or > internally on your own servers. > 4. Feedback on the usability of our EPUB 3 reader. > 5. Any other suggestions or comments that you think are relevant. > > Conclusion > ---------- > EPUB 3 with media overlays has a lot of potential, in particular in the education setting. Producing such books, however, is a > complex process. We believe that our system reduces this complexity to a level where many more organizations will find it > feasible to produce such books. Therefore, we hope for a positive response from the community. > > > > --- > You are currently subscribed to technical-developments as: janina@rednote.net. > To unsubscribe click here: http://cts.dundee.net/u?id=96295955.a9b72dc2d2019677a8bcc5371dd31c3a&n=T&l=technical-developments&o=6107866 > or send a blank email to leave-6107866-96295955.a9b72dc2d2019677a8bcc5371dd31c3a@mail.daisy.org -- Janina Sajka https://linkedin.com/in/jsajka Linux Foundation Fellow Executive Chair, Accessibility Workgroup: http://a11y.org The World Wide Web Consortium (W3C), Web Accessibility Initiative (WAI) Co-Chair, Accessible Platform Architectures http://www.w3.org/wai/apa
Received on Wednesday, 5 May 2021 12:33:31 UTC