Fewer than 100 lines of code for an initial recognition and synthesis application.

Fully configurable at the same time.

» Offline
» Easy to program
» Quick implementation
» Highly flexible configuration
» Free text recognition

» Multiple instances in parallel
» Use of multi CPU cores
» 65 different languages
» 145 multilingual voices
» Based on Java
» For Linux, Windows, ARM-CPU

Text-To-Speech (TTS)

» Multilingual
» 65 languages
» 145 voices (male/female)
» Plain text or SSML-Format

Automatic Speech Recognition (ASR)

» Fully event-driven
» Grammar-based
» „Barge-In“
» Gilded-Speech
» Low latency speech recognition (realtime factor 4,6%)
» Supports new acustic models

Use Case Examples

Keyword Recognition

“Which restaurant do you want me to find
for you?”

“Hmm, I feel like Chinese today”

Multilingual Speech Models

“Play ‚Clair de Lune‘ by Claude Debussy”