Rvoice Studio and Activeprompts

ActivePrompts are a new technology from Rhetorical, designed to offer a quicker and cheaper alternative to using voice talents and recording studios for the creation of an application-specific prompt library. Most speech user interfaces optimize the quality of their speech output by using pre-recorded prompts. In these applications, TTS is often considered to be a fall-back technology to generate speech for the variable content of messages. Although pre-recorded prompts optimize the quality of the static part of messages, they are expensive to create, depend on the availability of a speaker and recording studio, and can be difficult to combine with TTS. As an alternative to pre-recorded prompts, Rhetorical Systems has developed the technology to create prompts by using the rVoice TTS engine. This technology exploits the fact that rVoice contains a large unit selection database, that can be used to synthesize a large number of alternative versions of a same sentence. Given an appropriate interface to rVoice, it is generally possible to find the sequence of speech units that result in synthetic speech that can pass as a pre-recorded prompt. We coined the name ’ActivePrompt’ for prompts that are created from the rVoice speech unit database. They are different from recorded prompts in the way they can be used in the application. Instead of being a speech file, that is concatenated with speech that is synthesized, they can be stored as a sequence of speech unit identifiers that should be used to synthesize a particular text in a particular context.