At this time Computational Magic is still in an early beta version, which means many features are planned but not yet implemented.

Major features

  • Self service portal to create API keys, monitor API calls etc

Models currently in development

  • Image-Text extraction
    • A new model to find and extract text from image data, from anything like road signs, photos, scans etc;
    • (currently in internal testing)
  • Ichograph:
    • A transformer based, high accuracy automatic speech recognition (ASR) model that can transcribe spoken words to text
    • (currently in internal testing)
  • Text to Speech
    • Want your app to speak, read text to the user or even build a full virtual assistant? We want to make naturally sounding voices more accessible with a new text to speech model.
    • (currently in development)
  • Transformer Image Classification Model
    • Model to classify images, trained with a broad dataset to be able to perform many domain independent image recognition tasks (some industry / domain specific versions are also planned)
    • (in planning phase)
  • Transformer many-to-many language Translation
    • A transformer based model that can translate multiple languages between each other without intermediary. Right now in scope are German, English, Ukrainian, French, Spanish and Italian, but a lot more are planned to be added. Work on this is still early, so we can't give a planned release date yet.