IBM gives voice to the rise of Linux with ViaVoice

IBM gives voice to the rise of Linux with ViaVoice<@VM>Some of these 9 apps offer voice control for Web surfing

• Jun 21, 2000

The company's dictation software is the latest in a line of new categories of Linux applications

By Mark A. Kellner

Special to GCN

The dramatic rise of Linux as a rival to Microsoft Windows'in potential if not yet in practice for most PC users'has led to many applications being ported over to the open-source operating system. The latest new category for Linux is in speech recognition.

IBM Corp. expected to announce late this month the release of its ViaVoice dictation software for Linux. The move came a week after company officials said it would make its ThinkPad notebook PCs available with a Linux OS in the third quarter of this year.

ViaVoice, priced at $59.95, is billed as a full dictation editor, with which users can 'build documents with voice commands, do creation and editing with voice,' said Tom Houy, manager of client systems marketing for IBM Voice Systems in West Palm Beach, Fla.

In addition, users can copy and paste dictated text to other text editors or e-mail programs, just as they can with the Windows version of ViaVoice, Houy said.

Topics of speech

Array microphone: A flat microphone that sits on top of your monitor or desktop PC; eliminates the need for a headset in dictation. Continuous speech: The algorithm method used by most speech recognition programs to allow for continuous speech. Commands and punctuation need to be spoken'for example, 'period' or 'new line.' Recognition of text can be more fluid. Discrete speech: An early form of speech recognition, which required users to speak in distinct words, pausing between each. Enrollment: The process of training software to recognize your voice. Usually involves reading a prepared text. Programs also can scan your documents and e-mail to develop vocabulary lists of frequently used words. Normal mode: Continuous speech dictation mode lets copy be created with voice. Spelling mode: Useful when entering unique words or proper names. Most recent programs allow on-the-fly switching between modes with a voice command.

IBM said it has offered several voice elements for Linux over the past year'mostly in the form of software developer kits that can be downloaded free for personal use or licensed for redistribution. The company will now focus on individual users as well.

'IBM is very supportive of the whole Linux movement,' Houy said. 'We are bringing up every single software product in Linux. We've been very successful, and I think the Linux community has embraced IBM also.'

That may be an understatement: In the past 12 months, IBM has aggressively moved its product line into the Linux space. Although Compaq Computer Corp. is reported to be the leader in shipments of straight Linux servers, IBM appears determined to bring an enterprise focus to Linux computing, and that apparently means delivering a wide range of applications, including voice.

'We've been involved with doing the Linux stuff since about last April,' Houy said. 'The market is now light years ahead of what was available [then], both from software and hardware perspective.'

As with other aspects of the company's ViaVoice development, Houy affirmed that putting speech recognition on a desktop software platform is but one element of IBM's plans.

He hinted at the prospect of an additional Linux speech demonstration at the September 2000 GoMobile conference sponsored by Mobile Insights Inc. Visit the company's Web site, at www.mobileinsights.com.

Can we talk?

'Voice is one of key interfaces for the Linux community,' Houy said. 'It's going to expand in desktop and embedded systems.' He said that users of handheld devices will 'start going to Internet, and speech makes the mobile Internet possible; voice will be the interface of choice.'

The other major player in speech recognition, Lernout & Hauspie Inc., is taking a somewhat more cautious approach to the Linux platform.

L&H product manager Hank Pokigo said the company has its own speech software development kit for Linux, but is not yet creating a version of its popular Voice Xpress software for the platform. He said the company will watch the market and see if such a move is worth making.

Vendor	Product	Main functions	Platform compatibility	System requirements	Features	Price
Dragon Systems Inc. Newton, Mass. 617-965-5200 www.dragonsystems.com	Dragon NaturallySpeaking Professional 4.0	High-accuracy text dictation into any Windows application, edit text by voice, Web browsing, control of Windows desktop, text-to-speech, multiple user support, support of approved dictation recorders	Win9x, NT	Pentium; 48M of RAM for Win9x, 64M for NT; 200M of free storage; Sound Blaster or equivalent card	Has command Wizard for Macro creation, vocabulary of 250,000 terms and 160,000 words, formatting for Microsoft Word, Corel WordPerfect, GoldMine 4.0, Outlook, WordPad and Chat	$695
	Dragon NaturallySpeaking Legal Suite	Same functions as Professional edition, but features thousands of legal terms, precedents, phrases and abbreviations	Win9x, NT	300-MHz Pentium MMX, 128M of RAM, Sound Blaster or equivalent card	Includes terms designed for use in legal documents	$995
	Dragon NaturallySpeaking Mobile Organizer	High-accuracy text dictation into any Windows application, edit text by voice, Web browsing, control of Windows desktop, text-to-speech, control of PIMs	Win9x, NT	200-MHz Pentium MMX, 128M of RAM, Sound Blaster or equivalent card	Supports users of Microsoft Outlook, ACT, GoldMine, Palm desktop and Lotus Notes	$399
	Dragon NaturallySpeaking Preferred	High-accuracy text dictation into any Windows application, edit text by voice, Web browsing, control of Windows desktop, text-to-speech	Win9x, NT	200-MHz Pentium MMX, 128M of RAM, Sound Blaster or equivalent card	Basic business-level speech recognition program	$199
IBM Voice Systems West Palm Beach, Fla. 800-426-4968 www-4.ibm.com/ software/speech	ViaVoice Pro Millennium Edition	Command, control, and navigation of the desktop PC, provides Voice Mouse for hands-free mouse functions, takes direct dictation into Windows applications that accept text, provides voice- controlled Web browsing	Win9x, NT	233-MHz Pentium, 48M of RAM, 310M of free storage, Sound Blaster or equivalent card, Service Pack 4 for NT	Voice commands for desktop and system navigation; vocabularies include legal, medical and others; has a text-to-speech feature for proofreading	$179
	ViaVoice Pro Millennium Edition Elite	Same, but adds a flat array microphone	Win9x, NT	Same	Array microphone lies on desk or monitor, picks up voice while filtering out extraneous noise	$249
	ViaVoice for Macintosh	Only continuous speech dictation program for the Mac; text created can be copied into Mac word processors and other applications	Mac OS 8.5.1 and higher	G3, 48M of RAM, 200M of free storage, audio input jack for microphone	Creates documents, including e-mail, understands simple, natural, continuous speech	$89
Lernout & Hauspie Inc. Burlington, Mass. 781-203-5000 www.lhsl.com	L&H Voice Xpress Professional 5.0	Direct dictation into most popular applications, features Web browsing by voice command and speech-composed e-mail	Win9x, NT, Win 2000	266-MHz Pentium MMX; 64M of RAM for Win9x, 96M for NT and Win 2000; 250M of free storage; Sound Blaster or equivalent card	Type By Talking feature records up to 160 words a minute; accepts natural language commands, and has a 307,000-word vocabulary	$149
	L&H Voice Xpress Advanced 5.0	Direct dictation into many popular applications, features Web browsing by voice command and speech-composed e-mail	Same	Same	Shares many features with Professional edition but lacks advanced macro recording, voice tools and support for some applications	$89