Innovations in FonixTalk 6.1

Articulatory-vocal tract model

FonixTalk 6.1The application of a model that mimics the operation of the vocal tract and the vocal articulators is a significant step forward in TTS technology. This model gives a computer the ability to speak more like a human speaks and provides the flexibility to create the dynamic and varied speech sounds that make human speech unique.




What articulatory-vocal tract modeling provides

  • More natural sound speech that sounds less robotic.
  • Speech with quality similar to Concatenated Speech Technology that is based on recorded speech, while maintaining low memory and CPU usage by the integer version developed by Fonix engineers.
  • The ability to tune and customize voices.
  • Improved female and child voices.
  • The ability increase the speaking rate beyond normal speech while maintaining correct articulation and intelligibility.
  • The ability to create custom voices that match a real voice requires a fraction of the time and cost of concatenated voices Improved intonation modeling makes speech prosody (rhythm and stress) sound more natural.

Speech tuning tools

What developers and clinicians can do

  • Tune speech sound for specific audio channel or acoustic environment.
  • Easily create customized voices by adjusting model parameters.
  • Tune for a custom voice.
  • Tune voice for specific market needs, such as the Assistive Market.

Additional advancements

  • Voices utilize less than 1 KB of memory (concatenated systems require 100 to 1000 times as much memory.
  • Developers can add codes for added 'expressability.'
  • Significant code and algorithm optimization allow greater functionality to run on the same embedded platforms as previous versions of Fonix DECtalk, without significantly increasing memory and CPU usage.
  • New Unicode support of Asian languages.
  • New languages:
    • Chinese
    • Korean
    • Japanese in Q1 2008