Automated Speech Recognition & Speech Synthesis

ASTi's Speech Recognition software package option (ACE-SR) allows seamless integration with simulated software communication models for a one-box solution meeting the FAA/JAA (EASA) standard for flight simulators. The standard speech recognition package provides the ability to process up to 4 streams of audio simultaneously (depending on purchased options). Expansion beyond 4 streams is available as a special option. Speech recognition is configurable for word-by-word recognition to quickly recognize commands for system control or sentence-based recognition for use in "natural language" applications such as radio communications.

Potential applications include:

  • Instructor workstation voice control
  • Cockpit systems voice control
  • Automated Air Traffic Control and radio environment simulation
  • Talk-on target systems

The communication model contains a speech feed component while the speech recognition configuration file and grammar files reside in the Telestra project. Grammar files define the words for recognition as well as an associated result value for each word.

The speech recognition solution features the following:

  • Speaker independent (no training required)
  • Continuous recognition (phrases and sentences are both understood)
  • Background noise robustness
  • Dynamic grammar switching
  • Multi-channel support
  • Recognition success rates in the high 90's percentile
  • Multiple grammar formats available to simplify grammar specification