Speech Synthesis & Recognition

Overview Those of you who are familiar with the popular science fiction TV series “Star Trek” have already been introduced to the basic concepts of speech synthesis and voice recognition. On that TV series, characters spoke to the central computer and the computer spoke back. Today, however, most voice recognition software can only take dictation. It cannot comprehend words or their meaning. The ability to speak into a microphone and have your PC generate accurate text or recognize voice commands is the goal of current voice recognition software. Speech synthesis is the creation of speech from text through the use of your PC. Many new developments in digital speech technology, such as universal translators and truly dynamic voice recognition software, may be just on the horizon.

General Information Speech synthesis and voice recognition are both relatively new digital sound technologies. Until recently, digital speech tools were prohibitively expensive. Communications services have used this technology for directory services as well as e-mail and voice mail systems. Affordable speech synthesis and recognition tools are now available to anyone with a PC and a microphone. What is the difference between speech recognition and speech synthesis? Visit the Webopedia.com site to read their definitions of “voice recognition” and “speech synthesis”. Links to additional digital sound information can also found on these pages. Another good source of information about digital speech is the Speech and Sound Links page. Here you can find links to sites that deal with speech synthesis and recognition, sound analysis, and digital audio formats. An extensive list of FAQs is available at www.speech.cs.cmu.edu/comp.speech. Digital speech software tools, complete with descriptions and links to the manufacturers’ home pages, are available for download at www.speechandhearing.net/laboratory/tools.html.

Speech Synthesis Resources Synthesized voice of the past sounded mechanical and unnatural. Today, speech synthesis sounds more natural, due to the wide choice of voice types, cadence, and accents. Speech synthesis information, including more than 500 links, is available at www.speech.cs.cmu.edu. Links to speech synthesis and recognition resources can be found on this Speech Web Sites page. Speech synthesis utilities can be downloaded from this Web page. You’ll find both shareware and freeware programs included there. ModelTalker synthesizer, a modern speech synthesis program developed by the Speech Research Lab, is available for download at www.asel.udel.edu/speech/ModelTalker.html. Samples of synthesized speech generated by ModelTalker are available on this page as well. These synthesized speech samples will provide you with examples of the different features to be considered when using speech synthesis software.

Speech Recognition Resources Speech recognition software was first developed for PCs to perform tasks that include word processing and the execution of simple commands. Early versions of these programs were clumsy and unreliable. With each update in this type of software, there is an improvement in the accuracy of dictation. Newer software programs have more leeway for different accents, speeds, and voices. To browse resources, including FAQs, links, job postings, and discussion forums, visit the Commercial Speech Recognition page at www.speechstudio.com/commercial/speech.htm. Another broad-ranging resource is the Yahoo! Voice Recognition page. Here you can find links to software, articles, and tutorials. Speech recognition forums for both PC and Mac users can be accessed at this VoiceRecognition.com page. If you are looking for software that can add voice recognition capabilities to your word processor, then visit or www.nuance.com/naturallyspeaking.

Additional Links What is the future of speech synthesis and voice recognition technology? Visit HowStuffWorks.com to learn about a universal language translator that is being developed. Additionally, you can read about and listen to AT&T Labs’ speech synthesis software “Natural Voices”--some of the most naturally sounding synthetic speech.