INLOQ Voice Solutions

Traditional TTS systems are conceived to read texts in a single language. Multilingual texts can be read correctly by switching voices at every language change. However, this approach does not provide the best results when dealing with truly mixed-language text - where changes occur frequently and are embedded in sentences and phrases, as is the case of Web content, e-mail or information services, where foreign names and phrases (e.g. film titles) occur frequently.

The optimal solution is to have a single TTS voice capable of correctly reading an entire mixed-language text.

This can be achieved through two different approaches:

1. Producing multilingual vocal databases created with bilingual or multilingual speakers capable of reading several languages with mother tongue quality. For example, Loquendo's Jorge (Castilian) and Jordi (Catalan) in the Interactive TTS Demo section
2. Applying the phonetic transcriber to the foreign language text and then mapping the transcribed phonemes onto those of the voice's native language in order to access its acoustic units. In order to do this, the voice's acoustic database with the phonetic sounds belonging to its foreign language

While the first approach makes it possible to obtain "perfect pronunciation", it is however restricted to a small number of languages for a signle voice, whereas the second provides an approximate pronounciation, which is applicable to any foreign language. This doesn't mean that the first method is better than the second; on the contrary, in many real-life situations, the approximate approach is the most realistic. For instance, it is easier to understand an Italian film title pronounced by an English voice capable of correctly pronouning Italian rather than a mother-tongue Italian voice.