Browse Prior Art Database

Voicemail with Enhanced Speech Recognition: NLS Extensions

IP.com Disclosure Number: IPCOM000181579D
Original Publication Date: 2009-Apr-06
Included in the Prior Art Database: 2009-Apr-06
Document File: 2 page(s) / 37K

Publishing Venue

IBM

Abstract

Due to the widespread and aggressive globalization of business, many people from different locations often need to collaborate on a professional level. Unfortunately, not everyone speaks the same language. Employees all over the world are not necessarily bi-lingual or multi-lingual and those that are may not be fluent in all the languages. Clearly, language barriers are a common and inevitable challenge of a business - for both customers and employees alike.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 1 of 2

Voicemail with Enhanced Speech Recognition: NLS Extensions

Common language barriers faced by the hearing community have many parallel aspects to the challenges endured by the Hard of Hearing (HOH) and Deaf community. Disclosed is voicemail system architecture model capitalizing the customized voice user profiles and near 100% accurate voice-to-text translation (up to 160 words per minute) of phone calls, with a further extension of embedding a text-to-text NLS language translation facility feature.

     The invention is the incorporation of the language-to-language translation modules or software into a service-oriented architecture model so that someone speaking in native Mandarin Chinese would have their voice first translated into Mandarin text (recall this occurs at a rate of at least 160 wpm with 99% accuracy), and then second to the target language as text (e.g. English). The resultant translated text (in the target language) is either returned as a service or routed to/through the destination caller's voice translation profile, which might be either rendered to the display or perhaps reconstituted to audible English. Foreign callers can be identified via caller id or VoIP as the key that can be used for finding the matching caller's profile (via SOA services). This profile is then used by the NLS translation software.

     Virtually any example of impromptu use of this invention includes the need to pick up a phone to leave some message in one language, but have that message delivered in a second target language. Suppose a US travel company offers a message service where they can communicate messages on behalf of a traveler to some client that traveler may be attempting to meet, such as an arrival time or arrival delay. Suppose one of their benefits is that they offer this service with delivery within a 4-hour window. This service company must employ people round the clock or in different time zones to make this window. This service company could employ people living in BRIC countries to perform this work with the invention, where the employee would speak the message in a native Mandarin dialect which would get converted into a native English text (or audio) message by the time it got to the client.

Using the answering/voicemail scenario as an illustration:
I. Mandarin speaking caller (id#1234) dictates a recorded native Mandarin audio message intended for English speaking recipient (id#5678)

II. Inline telephone PBX (or exchange) or Web Service, dynamically processes the recorded message converting native Mandarin to Mandarin Text.

III. Mandarin Text is converted to English language Text using an utility like national language translator so...