The definition of Localization is to gather, collect, or concentrate in one locality. Localization or localisation can be defined as the process of translating a product into different languages, different customer segments or adapting a product for a specific country or region.

For IVRs, the difference between translation and localization is that you should not only take into account the target language, but also the target region for the language. Spanish, for example, has several variants, such as the Spanish spoken in Spain, the Spanish spoken in Mexico, the Spanish spoken in Cuba, etc. Furthermore, "localization" implies a deeper knowledge of the target culture. For example, a translator based in the United States may be fluent in the target language, but may not have the required depth of cultural knowledge around the target culture, to be able to provide intelligent edits to the design script. For example, knowledge of slang terms or idioms, any local knowledge specific to the country, such as transportation systems, geography, or history.

Also, note that VUI designs may have to be localized from one culture to another even if the source language and the target language remain the same. For example, for US English to Australian English.

When localizing an IVR system, you should solicit input from the local expert on the following relevant points:
  • Possible preference in the target region for gender-preferences. For example, is the knowledge domain seen as gendered? Should you possibly change the gender of the voice talent for the target culture?
  • Regional accents, or dialects. Are there certain accents in the target culture which will make your voice system sound like a different "persona," in other words, more educated vs. less educated, more rural vs. more urban, etc.

Conversely, you should provide the local expert input on any specific human-computer interaction expertise which you have brought to the design which may require special localization or translation. For example, you may wish to explain why you have constructed a given menu in a certain way, or the order of things in a playback format, so that the expert can take that into account when creating the translation.

Native speakers should always be used when translating an IVR script. Preferably, you should use not only a native speaker of the language but also a person who is familiar with the target culture, for localization purposes.