Speech-to-Text Gives Higher Education a Global Boost
Speech-to-text is making inroads into institutional learning. As the economy drives schools to go global, institutions of higher education are looking worldwide for revenue. Schools are spreading their reach into the global communities at home, as well as opening their classrooms for global broadcast as pay-per-course, and attempting to attract full-tuition students from overseas to travel to the United States for a degreed education.
The landscape of education is morphing in an attempt to overcome language barriers and seek revenue from countries and communities whose English skills may not be college level. Until now, language has been an insurmountable barrier, holding back students worldwide. If an American wanted to study in France, it would take only 10 minutes there to realize that five years of middle- and high-school French classes are not enough to prepare for the fluent, fast language spoken ad lib in a university lecture. The alternative to learning the language is course translation, but that is expensive. Professional translation (subtitles or voice) costs a minimum of $20,000 per course per language.
Through a combination of technologies, true globalization of education is not only possible, but expected. Speech-to-text can be used to recognize the spoken word of the original lecturer, or recognize the words of a "parrot" in the original language. A transcription can be generated infinitely faster than with manual methods, with vast time and cost savings. Speech-to-text can recognize the words of an interpreter, automatically typing directly into the new language.
Once a transcript is available, other software will turn these transcripts into correctly timed subtitles. Such software can be used to produce ready-to-view Web subtitles or courses burned onto DVDs for distribution.
Institutions also find the need to move their classrooms to mobile. For many students abroad, mobile and tablet may be the only reliable Internet access available. For others, they are simply preferred. Some companies have begun preparing mobile applications for internationalization and streaming courses. Intelatext.com not only displays subtitles on mobile, but also provides mobile software for teachers to converse with students across language barriers.
Mobile marketing technologists are being drawn into the globalization of institutions for several reasons. First, mobile is on the cutting edge of new content delivery. Second, it is in the strongest position to meet an institution's need to market itself in other languages. Third, it empowers universities to create and implement major mobile marketing campaigns—in many languages—at less cost and with greater success than with traditional methods.
When prospective students respond to a mobile marketing blast, recruiters can hold conversations across languages, using apps in which each side talks or types in their own language, each simultaneously translated and served up to a variety of delivery platforms.
In a variation on traditional marketing, companies like iXplore.com build online institutional marketing brochure templates compatible with do-it-yourself translation interfaces. These may be manually or automatically translated, then edited in an online environment, avoiding desktop publishing, printing, and mailing. These brochures incorporate subtitles over promotional videos, and the transcripts for those videos may be created using speech-to-text followed by human review.
Whenever languages and technology are involved, quality—of speech-to-text and of automated translation—is an issue. For live conversations across languages, such as a recruiter speaking to parents of a prospective student via Web or text conference, the conversational vocabulary tends to be more simplified, and thus speech-to-text and automated translation have greater potential for accuracy. Also, recruiters will take the time not only to use speech-to-text properly, but also to formulate their conversation for ESL (English as a second language) listeners. Speaking in an ESL pattern increases automated translation accuracy as sentences are less complex and contain the vital subject-verb-direct object often lost in natural speech.
To ensure quality for both speech-to-text and automated translation for courses and curricula, a clean-up process is mandatory. It must be performed by someone who understands the subject matter, as many university courses contain industry-specific vocabulary. But as long as clean-up is performed on the transcription (speech-to-text) and then again on the automated translation, the results come in at an approximate 90 percent reduction in cost over professional translation and a 50 percent reduction over professional transcription. That's a plus in any language.
Sue Ellen Reager is CEO of @International Services, a language and software solutions company that performs translation, voice recording, and global system testing for speech and DTMF applications, as well as media localization.
Middlesex County College benefits from a voice-based call attendant solution.