Browse Prior Art Database

Determining Theft of Text-to-Speech components

IP.com Disclosure Number: IPCOM000203278D
Publication Date: 2011-Jan-21
Document File: 1 page(s) / 19K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method to automatically determine and provide an audio-announcement of the origin and copyright status of a text-to-speech product.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 65% of the total text.

Page 01 of 1

Determining Theft of Text-to-Speech components

Unauthorized use of a company's Text-To-Speech (TTS) products can cost that company money. Source code, compiled code and language data-sets can all be mis-appropriated, pirated or re-distributed. When embedded into a product, it is virtually impossible to discern its origins because it lacks any visible manifestation. A means of determining the pedigree of TTS is desired. This invention provides such a means.

TTS works by streaming text, including both emotive declarations and linguistic content, through a processing unit which compares the text to a pre-defined data set, which then provides or can be used to derive the acoustic information necessary to generate comprehensible speech. The basic idea for allowing discovery of a pedigree is to hard-code a special sequence of characters into the data set such that, when encountered in the text stream, it will cause a verify specific declaration of pedigree to be verbalized. For example, if the text stream is "6%#asdf$a)))", the device's audio output is the product name and copyright information. Alternatively and more subtly, to prevent discovery and circumvention, syntactically invalid or highly unusual sequences of words or phonemes could be used to generate the foregoing pronouncement.

The invention works as follows:
1. Examine incoming text stream for key sequences. If matched, go to step 2. Otherwise, continue normal TTS processing.

2. Obtain or generate waveform...