System and method for rapid prototyping of existing speech recognition solutions in different languages
Inventors
KOPPARAPU, Sunil Kumar • SHEIKH, Imran Ahmed • PHARANDE, Amol Sitaram
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-8498857-B2
Publication Date
2013-07-30
Expiration Date
Abstract
A system and method for porting of existing speech recognition solutions in a source language to a target language has been disclosed. The system envisaged by the present invention enables porting of a working speech recognition solution in the source language to a working system in the target language, thus minimizing the development process and reusing existing speech recognition solution components to recognize multiple languages.
Core Innovation
A system porting a speech recognition solution in a source language to recognize a target language is described, while the speech recognition solution consists of a speech recognition engine, pronunciation lexicon in the source language, speech grammar file for the source language, and prompts in the source language. The system converts the pronunciation lexicon of the source language into an equivalent lexicon to be used in the target language by translation, transliteration, grapheme to phoneme conversion, and lookup table creation.
The conversion further uses grammar conversion means to modify the speech grammar file of the source language to handle free speech based speech recognition solutions in the target language by translating the grammar file to the target language and transliterating the translated grammar file to the source language to provide a transliterated grammar file for the target language. In parallel, prompt generation means convert voice prompts in the source language to the target language by using identification means with the lookup table, text to speech conversion means, grammar adjustment means, and speech generation means to provide speech based output for a grammatically modified text based prompt.
Prompting means output the generated speech to guide users to submit their query, receiving means receive a speech based query from users in the target language, and processing, compiling, and playback means provide a final output in the target language. The independent claims cover converting a source-language speech recognition solution into an equivalent target-language solution by lexicon conversion, grammar conversion, and prompt generation, with the system claim further specifying end-to-end target-language operation.
Claims Coverage
The document provides two independent claims, a system claim and a method claim, each focused on converting a source-language speech recognition solution into an equivalent target-language solution. Across the independent claims, three inventive features are emphasized: lexicon conversion, grammar conversion, and prompt generation, with the system claim additionally covering end-to-end receive/process/compile/playback in the target language.
Lexicon conversion means for equivalent target-language lexicon
Lexicon conversion means convert the pronunciation lexicon of the source language to an equivalent lexicon to be used in the target language, using a first database for analogous source- and target-language words, translation means, transliteration means mapping translated words into source language graphemes, grapheme to phoneme conversion means generating source language phoneme sequences, and lookup table creation means preparing a lookup table.
Grammar conversion means for free speech based target-language grammar
Grammar conversion means modify the speech grammar file of the source language to handle free speech based speech recognition solutions in the target language, including translation means to translate the speech grammar file to the target language and transliteration means to transliterate the translated grammar file to the source language to provide a transliterated grammar file for the target language.
Prompt generation means for grammatically modified target-language speech prompts
Prompt generation means convert voice prompts in the source language to the target language using translation means, identification means using the lookup table, text to speech conversion means, grammar adjustment means arranging the words in accordance with the transliterated grammar file, and speech generation means generating speech based output for the grammatically modified text based prompt.
Target-language end-to-end operation using converted source-language representation
Prompting means output generated speech to guide users, receiving means receive a speech based query from users in the target language, processing means perform a predetermined operation on the source language text, compiling means compile the final output in the target language using lexicon conversion means, speech grammar conversion means, prompt generation means and processing means, and playback means play the final output in the target language.
Overall, the independent claims cover converting a source-language speech recognition solution into an equivalent target-language solution by lexicon conversion, grammar conversion, and prompt generation. The system claim further specifies target-language prompting, receiving target-language speech queries, converting them into source language text and source language representation, compiling the final output in the target language, and playing it back.
Stated Advantages
Reduced porting effort/time by enabling reuse of a speech recognition engine and by generating target-language resources using conversion and lookup table techniques.
Documented Applications
Porting an IVR-style speech recognition application from a source language to a target language while reusing unchanged call-flow units/business logic and keeping textual processing unchanged.
A described example of porting English to Hindi as part of the conversion workflow.
Interested in licensing this patent?