GB2422449A

GB2422449A - Text to sign language translation software for PCs and embedded platforms e.g. mobile phones and ATMs.

Info

Publication number: GB2422449A
Application number: GB0501106A
Authority: GB
Inventors: Christopher David Taylor
Original assignee: Individual
Current assignee: Individual
Priority date: 2005-01-20
Filing date: 2005-01-20
Publication date: 2006-07-26
Also published as: GB0501106D0

Abstract

A program for converting the written word and to Sign language for implementation on x86 Personal Computer and embedded devices (mobile phones, ATM machines etc...). Including a means of generating Sign language videos/animation with subtitles for educational use. It provides a means to save videos/animations to hard disk and use the software as a quick reference guide for commonly used expressions and as a dictionary. A sample version is working and the user interface for x86 platforms is displayed in Fig 1. The concept of extending the program to convert text messages and voice mail to Sign language is conceived.

Description

Sgri Studio This corcept relates to the the translation of written text to Sign language via computer sDftware.

There are a number of scenarios in daily life in which the ability to quickly learn Sign language is desirable. Or there is the possibility that one may be thrown into the situation vhere even t'iough they have little training in the matter, they need a reliable learning ntera to issist them in quickly applying signs.

FEowever, it can be an extremely time consuming process (hours to translate about a page of text) and still images do not always clearly illustrate how a Sign should be applied, so in an ideal case, video clips or animations would be used to represent Signs. There are also a number of other scenarios where it is desirable to quickly learn sign language and have a quick reference material I) Staff in a care home may be faced with an emergency placement where a hearing' irnpired individual has to be placed in their care, It would be desirable to be able to Sign to them, but the staff have very little Signing experience. Only simple expressions are required but still the ability to convert customized text to Sign would be desirable.

2' Students of courses to learn Sign language require a quick reference material to convert mass blocks of text to Sign. It would take seconds to convert the text to sign as opposed to hours currently. The ability to quickly look up Signs for individual words would also be an advantage over current means.

3) An individual who has lost hearing, for example due to meningitis would suddenly need to learn Sign language very quickly. Family and friends of such an individual would also find it adva ntageous.

Referring to Fig 1. where the demonstration program is used, text may be entered at (I).

Th er is notified of any unrecognized words by means of a list at(2). The software plays back the Sign in the media player (3) a means is available by which the user may control the playback rate (4).

Subtitles are added to each sign so it is clear which word is being represented by a specific sign. The rate at which the Sign video/animation is played back may be adjusted so that warners can nore easily identify individual Signs, and how to perform them for themselves. It

The workings of the program and its operations will be described with flowchart Fig. 2 anc cipnyu notes below. Fig 2. displays a flowchart which governs the basic operations which the conceived program goes through in order to perform the conversion process.

Detailed information of the workings of the demonstration programs' operations is given below. This description is for the x86 version of the software.

1. Textual input is received; Textual input may be received via a number of methods, all dependent upon the implementation of the program. They could be; Stored in a text file s a paragiaph (there is no specific limit on the number of words in any given fife).

defined phrases which may be in Strings hard coded into the program. Another source could be text which was copied and pasted directly into the program from any external source such as a website.

2. The text is split down into Component words and punctuation using the String Processing Functions of whichever programming language the program is written in. This is in preparation for the next stage. Certain errors are caught here which could cause the program to function incorrectly. For example, in the text translator, if only 1 word been selected as part of the phrase, the program will halt the process.

3. Grammar that could cause the program to falsely identify words as spelled incorrectly is filiered out of the Strings. Important grammar is preserved by leaving an instruction to the program that there is a case where it is desirable to appropriately Sign grammar.

4. Spell checking is performed. The program searches its database of installed media for appropriate Signs for each word in the text. At this point if a word is spelled incorrectly cr not found. The user is informed that the words in question may be spelled incorrectly cr not found. He/she will then be advised to run web updates which checks for newly died words and installs them if required. Correctly identified words will be marked as cky to include in the rest of the conversion process.

5. A list of media files is generated containing all correctly identified words and grammar in tie appropriate order. Having selected the best possible match of Signs, at this point the computer will attempt to merge them together into one large media flle and play it had using an embedded player or an external one. This is largely dependent upon the Operating System that is used. 3 5

"he program is slightly different for embedded platforms. Fig 3. is a flow chart which oitlines the process according to the programs' implementation specifically on mobile phone type embedded platforms. More detail is given regarding the operations below. Note thai the process so similar that the below notes only refer to variations between this version and the x86 implementation on the previous page.

1' Textual information is received from an external source such as a web page, or text message. This may also include voice mail to text as a source but in this case, the voice to texi conversion would already have been done remotely. At this point the text is stored direcdy in a string.

2) The text is separated into an array of strings with grammar preserved using String Processing Functions. In this case, single word text messages are allowed.

3) Grammar which could cause falsely identifying words as spelled incorrectly is removed.

Furthermore, since memory and other performance aspects are limited on such platforms, the amount of grammar removed will increase with the limitations of the device.

4.' The. program can then request that a remote computer server checks which words are spelled correctly and have remotely stored media files.

5) The media files are sent back to the mobile/embedded device. The program will play bark each iedia file corresponding to a Sign in the correct order. No attempt will h rnicJe to atach each file to the other, since available memory is a constraint and the playback should be relatively smooth anyhow because no hard disk is usually present sc buffering of large clips as data is read off disk should not be an issue..

Claims

Claims Sign Studio I. The concept of computer software which converts

written language or textual information to Sign language for the hearing impaired, such that a video, animation or sequence 01' stil images is produced which accurately reflects the content of the initial message/paragraph. The program has the ability to add subtitles to signs if desired and r1v act as i dictionary and general phrase book for quick reference translating commonly used daily expressions.
2. The concept of implementing the software as outlined in claim 1 on x86 - 32/64 platforms on its own or as a component of a larger software package as an educational tool or ctherwise. Also including the ability to save Signed video/animation these platforms.
3. The concept of implementing the software outlined according to claim I on embedded platforms such as mobile phones and ATM machines, by which similar software may be used to co'ivert incoming textual messages to Sign language, with the accompanyin media tiles used to construct the Signed video/animation stored either remotely or locally
4. The concept of implementing the software outlined in claim 1 as described in claim 2 with additional functionality for Instant Messaging Clients, or as part of an Instant Messaging Client.
5. The concept of implementing the software outlined in claim I as described in claims 2 & 3 with the equirement that no specific proprietry file format or proprietry programming coi-nponent is required, such that existing documentation should suffice for recreating the So'iw are iii any programming language.
6. The concept of software as outlined in claim I implemented in any of the means described in Claims 2 to 5, where String Processing Functions are applied (at the programming level) to identify grammar, words, phrases and special language cases in the analysis of text in order to create a Signed video/animation.

1 Th concej't of implementing the software outlined in claim I in languages other than UK English and British Standard Sign on computer platforms as stated in claims 2 & 3.