Using a mathematical representation of the soundscape based on years of research and development by Prof. Kevin Short, our algorithms separate the speaker of interest and allow enhancing and remixing to achieve optimized input into speech recognition or communication systems.

Setem Signal Separation is based upon a mathematical decomposition of signal elements within the source signal(s), and enables the separation of targeted sound sources from interfering elements, which can then be reconstructed individually or in groups.

  • The core technology is a proprietary solution to the “Cocktail Party Problem” – enabling a receiver to focus on a single voice or conversation in a crowded audio environment.
  • The embedded algorithms can separate and enhance the desired audio source based on a variety of signal characteristics or user controls
  • Setem’s software processes the received signals to enable selective enhancement of speakers or conversations, while minimizing unwanted background noise.
  • Setem’s processing algorithms work in real time.
  • Setem Source Separation outputs either audio or processed signals.

Setem’s solution does not require any additional or proprietary chipsets / silicon – it can be integrated into existing smartphones using current microprocessors, microphones and other standard technology with minimal to no incremental power consumption.