Spoken Language System customization

Commercial products often do not perform well right out of the box, and either need customization or are outright inappropriate for the application, in which case one has to consider building a new system that matches the specifications of the application.  We work on these projects:
  • Application-dependent speech recognition
    • Acoustic model training for speech recognition
    • Dictionaries for ASR
  • Application-dependent speech synthesis
    • Mixed prompts and diphone synthesis
    • Special purpose text analysis
  • Small footprint versions of ASR and TTS
  • Altered versions for special applications, such as client/server speech synthesis