Zamia AI - Features

Speech

Automatic Speech Recognition

Zamia AI offers support and pre-built models for state of the art speech recognition engines such as Kaldi-ASR and CMU Sphinx.

Speech Synthesis

For speech output, Zamia AI has you covered through its future-proof TTS abstraction layer. This allows you to use any TTS engine you like - and comes with drivers for these TTS engines:

  • eSpeak NG
  • MaryTTS
  • SVOX PicoTTS

Language

Natural Language Understanding

Zamia AI offers advanced deep learning based natural language processing features:

  • Contexts
  • Semantics roles
  • NER
  • Relations
  • Temporal and event processing

Reasoning

Knowledge Base and Reasoning

The foundation of Zamia AI's knowledge base is XSB Prolog. This enables state of the art efficient reasoning.

Learning

One of the key architectural features of ZamiaAI is that it can learn on many layers:

  • NLP: Deep learning Seq2seq neural network
  • Python: Easy to use persistent value store
  • Prolog: Standard Prolog builtins such as assertz/2 can be used to modify and extend the knowledge base dynamically

Architecture