Using multilingual voices

Vocalizer for Enterprise supports a number of multilingual voices. A multilingual voice is one that can pronounce text in different languages.

For a list of multi-language voices, see Vocalizer languages and voices. To get details for any given voice, see the NuanceLingwareDocumentation that downloads with that voice.

To get details for any given voice, see the NuanceLingwareDocumentation that downloads with that voice.

Multilingual voices are useful where the text being converted to speech frequently uses foreign terms. For example, a US English multilingual voice with Canadian French capabilities can be used in a navigation system to pronounce place names correctly for drivers visiting Quebec.

A multilingual voice may be able to read several different languages, but with varying levels of accuracy. Each voice is rated for each language it can speak:

  • A superior rating means the voice is able to speak the foreign language with only the barest traces of an accent.
  • Enhanced means that the voice speaks the foreign language with a noticeable accent, but nonetheless accurately.
  • Standard means that the voice attempts to speak the foreign language by defining the words in phonemes of its own language. This can lead to inaccurate pronunciation when the phonemes of the foreign language are not present in the voice language.
  • Basic means that the voice simply reads the foreign language text in its own language.

For example, the German voice Anna-ML is superior in French, Italian, English, and Spanish, but only basic in Dutch.

When the voice is rated superior or enhanced for a language, the resulting speech is easily understandable to a native speaker of the foreign language. A standard or basic language still reads text marked for that language, but often with far less accurate results.

Activating a foreign language

The foreign languages associated with a multilingual voice are activated automatically when you select and load the voice.

Marking up the text

By default, a multilingual voice reads text in its own language. To read a section of text in one of its supported foreign languages, you mark up that section using the control sequence:

<ESC>\lang=language\

You must use the escape sequence twice: once to indicate where the foreign language begins, and a second time to switch back to the original language. For example, to read a Spanish place name in a German multilingual voice:

Die Ausfahrt Richtung <ESC>\lang=SPE\ Palma de Mallorca <ESC>\lang=GED\ kommt nach 120 Metern.

This example says "Die Ausfahrt Richtung" in the default German. At the escape sequence, the voice switches over to the Spanish language to say "Palma de Mallorca." The next escape sequence switches the language back to German to say "kommt nach 120 Metern."

For voices that support multiple languages and multiple styles (-mls), you can mix both. This example speaks the first sentence as conversational English, the second as conversational Spanish, and the third as formal Spanish:

<esc>\style=conversational\ This is conversational text in English. <ESC>\lang=SPM\ This is conversational text in Spanish. <esc>\style=formal\ Thank you for listening.

See Changing the speaking style.

Identifying an unknown language

In cases where the language is not clear, Vocalizer for Enterprise can try to identify the language. If it is able to do so, it pronounces the marked text accordingly.

Use <ESC>\lang=unknown\ to activate the automatic language identifier: 

Die Ausfahrt Richtung <ESC>\lang=unknown\ Palma de Mallorca <ESC>\lang=GED\ kommt nach 120 Metern.

If Vocalizer for Enterprise does not support an indicated language, or is unable to identify the language, it ignores the markup and reads the text in the voice's default language.