Techniques for aiding speech-to-speech translation

A computer program product comprising a tangible computer-readable recordable storage medium having computer readable program code, said computer readable program code adapted to be executed to implement a method for translation of a source language utterance of a first user, said source language utterance being in a source language, said first user being a speaker of said source language, said method comprising:

• providing a system, wherein the system comprises distinct software modules, each of the distinct software modules being embodied on said tangible computer-readable recordable storage medium, and wherein the distinct software modules comprise a speech recognition module, a machine translation module, an input/output module, and an information retrieval module;
• performing automatic speech recognition on said source language utterance to obtain a speech recognition result corresponding to said source language utterance, by executing said speech recognition module on at least one hardware processor;
• performing machine translation on said speech recognition result to obtain a translation result in a target language, by executing said machine translation module on said at least one hardware processor;
• performing information retrieval on a supplemental database, based on said speech recognition result, to obtain multiple retrieved word strings, in said source language, related to said source language utterance, said retrieved word strings having word string translations in said target language associated therewith, said supplemental database comprising a parallel corpus with indexed training data including training word strings in said source language and corresponding translated training word strings in said target language, said machine translation module being trained on said training data of said parallel corpus, said performing of said information retrieval comprising:
• taking said speech recognition result as a query in a weighted combination;
• and
• searching said indexed training data based on said query;
• said performing of said information retrieval being carried out by executing said information retrieval module on said at least one hardware processor;
• and
• formatting said speech recognition result and said retrieved word strings in said source language for display to facilitate an appropriate translation selection, by executing said input/output module on said at least one hardware processor.