IBM releases source code of speech software

Big Blue is releasing code to the Apache and Eclipse Foundations, but users will need to buy a proprietary speech-recognition system to take advantage of this

IBM will release the code of various runtime and tool components to the open-source community to make it easier for developers to add speech-recognition capability to Web applications. The company announced earlier this month that it will give software components to the Apache Software Foundation, an organization that supports a range of open-source software projects, to make it easier for Web developers to build speech-enabled Web applications.

It is also planning to give speech mark-up editors to the Eclipse Foundation, an open-source software development platform, to help developers to write standards-based speech applications.

Mike Milinkovich, the executive director of the Eclipse Foundation, told Builder UK on Friday that the mark-up editors will be implemented as an Eclipse plug-in and will be integrated with the other tools that it already provides. He said that increasing the number of speech-enabled applications will not just make it easier for people with disabilities to use a computer, but could eventually revolutionize the way we interact with computers.

"Voice is a more natural way to interact with computers. Accessibility is a key part of it, but the hope is to make voice applications truly ubiquitous," said Milinkovich.

Olaf Schmidt, a developer on the KDE Accessibility Project, which develops accessibility applications for the Linux desktop, told Builder UK that the components that IBM has released are useful for developing Web applications, but will not help with developing desktop applications. Schmidt also pointed out that people with disabilities who rely on speech-enabled applications need to have IBM's ViaVoice or some other speech recognition system installed to make use of the code donated by IBM.

The high cost of proprietary versions of accessibility software versions can often prohibitive, particularly as many people with disabilities are unemployed. This was pointed out by Janina Sajka, the chair of the Accessibility Workgroup of the Free Standards Group in an earlier interview.

First Linux-based speech app?
Schmidt said there is a wide range of accessibility software available for the Linux platform, but speech to text applications have not yet been developed. "There is no speech-recognition system available for Linux, which is a big gap. We need a release of IBM's ViaVoice with a license that allows us to use it with Linux-only free software."

Speech recognition software is essential for people who no longer have the full use of their hands, such as Multiple Sclerosis (MS) sufferers. "Some people with MS can only move one finger and can take a minute to type just one character," said Schmidt.

ViaVoice was made available for Linux in June 2000, but was later taken off the market. Aaron Leventhal, an IBM employee, said at a KDE conference in August that ViaVoice's text to speech engine would be made available for Linux through Wizzard Software. He made a similar statement in a presentation about accessibility to the developers of open-source browser Mozilla.

But if users wish to use this software, which does not include the speech-recognition components, they must develop their own application in a Windows environment and then compile it on Linux, according to the Wizzard Web site.

IBM was unable to provide a comment on this issue at the time of writing.

Another hope for Linux users who need speech-recognition software is Sphinx, an open-source speech recognition project. But according to the project's Web site this is not a final product, and Schmidt is not aware of any desktop project using it.

Ingrid Marson reports from ZDNET UK