DictationBridge: A Graduated Schedule


The DictationBridge crowdfunding campaign was launched in late April and, as of this writing, has raised approximately 26% of its $20,000 goal. The DictationBridge campaign uses the Indie Go-Go “flexible funding” model so the team will receive the dollars contributed to the effort even if we do not make our entire target.

The DictationBridge team has received a number of questions from individuals who want to know what we will do if we do not make the entire $20,000 goal.

It should be noted that each level is cumulative so everything that goes into a lower level of funding will also be in those that require a higher level of financing to accomplish.

$7500: The Basics

The DictationBridge team met and added up the costs to determine our minimum level to be able to produce a useful piece of software and we found that for $7500, we could develop the following:

  • Support for NVDA with Windows Speech Recognition (WSR).
  • An easy to translate package that others can bring to non-English languages.
  • Basic documentation describing how to install and use DictationBridge with WSR.

When we reach the $7500 milestone, the engineers will commence work on the real DictationBridge software (as opposed to the demo) released to the public. Those who joined the beta team by donating $100 or more will soon afterward start seeing incremental builds of what will be the released version of the software.

At this level, the single most challenging task is replacing the proprietary API hooking library used in the DB demo with the free and open source one used by NVDA itself.

$10,000: Dragon Comes To NVDA

At the $10,000 contribution level we will, in addition to the tasks at the previous levels:

  • Add support for the Dragon speech recognition from Nuance Communications to the main DB DLL.
  • Write NVDA scripts to support the Dragon user interface elements.
  • Expand the documentation to include instructions for using DictationBridge with the Dragon products.

This level has two relatively complicated tasks. Unlike WSR, the Dragon products use a variety of different APIs to insert text into a document, and while the DictationBridge demo already shows off Dragon working in edit and rich edit controls, we need to find which other API it uses for text in MS Word, in web controls and in other areas as well. The APIs used by Dragon aren’t documented publicly so the DLL related tasks will require Matt to poke around under the hood in Microsoft Windows to figure this out. The other time consuming task is writing the scripts to support the Dragon interface, including the screens used to train the Dragon software to better understand your speech patterns.

$12,500: Control NVDA with Speech Commands

At the $12,500 level, in addition to everything described above, we will:

  • Write Dragon Professional scripts to provide a way for DictationBridge users to issue spoken commands to operate their screen reader. We intend to include as many different NVDA commands as possible at this level.
  • Ensure that the Dragon Pro scripts are easy to translate into non-English languages.
  • Write the documentation detailing how a user can employ these commands when using DictationBridge.

At this level, the technical challenges aren’t too great as Dragon Pro scripts are fairly straight forward to write and provide a number of ways it can be used to communicate with the main DB DLL. These tasks, however, are time consuming and present a difficult set of test cases to determine if a bug is in the DB DLL, the Dragon Pro scripts, the NVDA Dragon scripts or NVDA itself. As there will be a large number of NVDA commands supported, this also creates a time consuming task for those writing the documentation.

$15,000: DictationBridge Comes To Window-Eyes and ZoomText Fusion

At the $15,000 level, in addition to everything described above, we will:

  • Write scripts for Window-Eyes (WE) and ZoomText Fusion (ZTF). It’s our understanding that this will be a single set of scripts as the scripting facilities in Window-Eyes and ZTF are identical.
  • Write the Dragon Professional scripts to allow for users to control WE and ZTF software with speech.
  • Ensure that these scripts are as easy to translate into non-English languages as possible.
  • Expand the documentation describing how to use WE and ZTF with DictationBridge.

The biggest challenges at this level are adding a second and third access technology to the package, writing the scripts for WE and ZTF, expanding the documentation and vastly expanding the test cases to ensure that DB is working properly with both NVDA and the two access technologies from AI Squared.

$17,500: DictationBridge Comes To JAWS

JAWS, according to the best data we have available today, remains the most popular screen reader in the world. But, as one can read in Matt Campbell’s description of the problem entitled, Inside DictationBridge, “the JAWS scripting facility is considerably less powerful than the more modern analogues available for NVDA and WE/ZTF. The biggest issue is that JAWS can query applications for information through any number of different techniques but it cannot be called by an external application so it’s impossible to notify JAWS that something has happened on the screen in an application that does not have focus.”

Both WSR and the Dragon products pop their UI onto the screen in a non-standard manner and leave the application in which the user had been working with focus. The only way to add a “global” script to JAWS requires altering the default scripts shipped with JAWS itself. The DictationBridge team, while it hopes to serve as many users as possible, has elected to not change the JAWS default behavior as doing so means that any changes made by Freedom Scientific or a third party (Doug Lee’s popular Skype scripts for instance) will require this part of the DB code to also be changed. Altering the default behavior of JAWS will also cause technical support issues that, based on any number of different factors, will be very hard to impossible to solve as using this technique will insert a large number of additional variables to the team’s test cases.

Thus, if we reach the $17,500 point, in addition to everything described in the sections above, we will:

  • Write JAWS scripts to support the Dragon line of products from Nuance.
  • Write Dragon Pro scripts to permit users to issue JAWS commands by voice.
  • Ensure that the scripts are easy to translate.
  • Expand the documentation to include how to use JAWS with DB and describe the limitations it has compared to the other screen readers we’re also supporting.

$20,000: The Gold Plated DictationBridge

If we reach our entire $20,000 goal, along with everything described above, we will:

  • Attempt to support WSR with JAWS.
  • Ensure the documentation is professionally written and that people have tested it for clarity.
  • Perform a higher level of professional level testing.
  • Possibly include features suggested from users “in the wild” that we on the DB team hadn’t thought up on our own.
  • Put a few dollars in the bank to pay developers to maintain DB if/when something like a new version of Windows, WSR or Dragon might cause failures.

What If We Fall Short Of The Minimum?

While the DictationBridge team is highly confident that we’ll reach the $7500 milestone, as of this writing, we had approximately $5200 in donations with roughly 40 days remaining in the campaign, so, although we think it’s highly unlikely that we will not reach the minimum, it is still a possibility. Thus, if we do not hit this milestone, we will:

  • Repay the debts we’ve incurred to get the dictation campaign launched. This amounts to approximately $2000 that we spent mostly to hire and train the technical support staff and to acquire some copies of Dragon, a copy of the 90 day test version of JAWS and a few other odds and ends. If we do not reach the minimum, nobody on the team will be compensated for their time and no additional expenses will be considered.
  • Donate the rest of the money (minus of course the Indie Go-Go fees) to the NVAccess Foundation to be used as Mick Curren and Jamie Teh, the amazing engineers behind NVDA, see fit. Nearly every member of the DictationBridge and 3 Mouse technology teams are NVDA users and we all trust that Mick and Jamie will use these dollars judiciously. NVDA is the only Windows screen reader to have shown marketshare growth in each of the past six years and is a tremendously valuable tool for blind people in nearly every corner of the globe. Obviously, our goal is to raise the entire $20,000 and deliver the gold plated version of DictationBridge as we believe it will add a powerful set of additional features to the different screen readers we intend to support but, if this proves impossible, we’re proud to help NVAccess in any other way we can.

In Summary

The DictationBridge team is committed to making as good a piece of software as the budget will allow. Fiscal realities, like being able to pay the developers, testers and those writing documentation have led us to create this collection of milestones so as the donating public can understand what each step will cost and what each milestone will allow us to build. Nearly everyone working on DictationBridge is a screen reader user so we all have a stake in its outcomes and we believe strongly that access technology for blind people should be designed and developed by members of our community and that blind technology professionals should be paid for efforts that benefit the entire community. This is a “for us, by us” project and we hope you choose to donate to the DictationBridge crowdfunding campaign before our June 20 deadline.