Text-To-Speech Experiment & Evaluation: Cognable Speeka

I created a test page for an experiment with Speeka text-to-speech (TTS), graciously provided by Simon Evans of Cognable. I plan to incorporate TTS into every page of the future Clear Helper Web site.

Background

Speeka, a free service, is a work in progress. It is not a polished, commercial product. It is one of many Mr. Evans is developing to improve accessibility for people with intellectual disabilities.  A brief description of each of his projects can be found on the Cognable home page.

I think Speeka’s initial implementation was on the the Web site of Inclusive New Media Design. INMD is an organization that, like me, is working to develop best practices of Web accessibility for people with intellectual disabilities. When I first saw Speeka, I immediately liked its small form factor compared to that of ccPlayer, which I have been using.

Appearance & Placement

Speeka is embedded throughout the INMD site in the top, right of the content section. It appears as the image below. Rectangular. 3 buttons: play, back, forward. A speaker symbol and the word 'listen'On my test page, it appears as the following image.3 buttons: play, back, forward. The words 'audio stopped' underneath

I too placed it in the top, right of the content section.  Of the Web sites I have visited that use a TTS feature, most embed it in a similar location.  Those that don’t place it on the bottom of their pages.

Configuration

Setting up Speeka in my test page was a simple affair.  I inserted the HTML code provided by Mr. Evans.  I needed only to change the referenced file name.  I made one addition; that of the application landmark role to Speeka’s container. This helps people with screen readers, who use WAIARIA, to identify it. Upon placing the test page on the Clear Helper Web site, I invoked a hyperlink Mr. Evans provided to inform Speeka of the page’s presence.

Voice-Narration

I configured Speeka so it reads only primary content.  It can be set up to read all the textual content of a page, including menus, but I suspect it would be tiring to listen to the same menu over and over.

I chose to use a natural sounding, British male voice. [Edit on 2010-01-31: The voice is now an American one.] The test page it is reading contains text written as simply as I could at the time. Its pronunciation of the words and the sentences is very good. It had no problem with my last name.  I will have to test it with more complex text and with unusual proper nouns.

It announces every heading with the word “heading”; each list item prefaced by the word “bullet”; and the beginning- and the end of every list.  I was surprised. This feature is the first I have experienced with a TTS application.   It may be useful, but I think it would better serve as an option. [Edit on 2010-03-14: Announcement of list bullets, beginnings and ends is now an option. It is not active on the test page.]

General Navigation

The three-button interface is simple.  The audio narration can be played and paused with the same button. The forward button advances the narration by six seconds; the back button rewinds it by four.  Suggestions:

  • Perhaps it would be better if the forward- and the back buttons advance and rewind to adjacent sentences.
  • An option to restart the narration from the beginning may be helpful.  The only way I could do it was by refreshing the page using the Web browser.
  • Audio- and visible text labels for the buttons are a necessary feature, I think. An example can be found in a BBC Flash Player designed for people with intellectual disabilities.  It can be seen on the BBC’s Us 5 site, by clicking the link “Launch Us5 videos in pop-up windows”, then by selecting an actor.

Keyboard Navigation

Pressing the Tab key cycles through the buttons. The Space Bar or the Enter key invokes them. I had no trouble with this navigation within Speeka, but I could not tab inside the Web page to get to it. I could use the Tab key with Speeka only after changing focus to it by clicking it with my mouse.  This is not unique to Speeka.  I experienced the same with ccPlayer.  Keyboard navigation is important because many people with intellectual disabilities also have physical ones.  Such disabilities often preclude the use of a mouse, and require keyboard use or a single-switch device.

Interface Text

When the play button is clicked, the “audio stopped” text changes to a countdown of time until the end of the audio narration.  I think being presented immediately with the “audio stopped” text is potentially confusing.  I also think both it and the countdown test may not be necessary.

Speaka-Service Functions

Speeka converts Web-page text to MP3 files.  When a Web site visitor clicks the play button, the MP3 is streamed to the visitor’s computer from a Cognable server.  This is advantageous for Web sites that do not have a streaming-media server nor the bandwidth to support one.

A great feature of Speeka is it checks the text of each page on a regular basis.  When it detects a change, it updates the associated MP3 file.  Graphed statistics about this can be found on the Speeka home page.

Conclusion

Speeka has many nice features.  I think its inclusion on a Web site designed for people with intellectual / cognitive disabilities would provide site visitors with a significant accessibility feature. With all of Mr. Evans’ projects, I don’t know if he has the time to consider some of the options I have mentioned, but I plan to discuss them with him.

Note: No endorsement of Speeka or Cognable is expressed or implied.

Tags: , ,


%d bloggers like this: