skip to main content


There is now a massive demand to "wake up" and control devices and systems by voice, specifically by the voices of people who have the right to issue commands, and this needs to be achieved in noisy environments. This "cocktail party" problem of tracking individual voices in a crowd has held back the widespread adoption of voice-controlled devices in the mobile environment.


  • Extra linguistic scene analysis AI is used to track, identify, and enhance a specific voice of interest to ensure that the signal-to-noise ratio (SNR) of the targeted voice is greater than 10 decibels (dB), a range over which state-of-the-art speech and voice recognition technologies can be effective.

Features and Benefits

  • Assesses the adaptive reasoning method that allows for effective operations with no a priori information about the direction or location of any of the voices and sounds contributing to the audio signal.
  • Learn how the voice control of devices can be enabled in high-noise environments, specifically in far-field environments where the noise sources are closer than the speaker.

Key questions answered

  • How does Yobe outperform the competition in solving the "cocktail party" problem in voice recognition?
  • How can enterprises have access to this technology?

Table of contents


  • Catalyst
  • Key messages
  • Ovum view

Recommendations for enterprises

  • Why put Yobe on your radar?


  • Background
  • Current position
  • Roadmap

Data sheet

  • Key facts


  • On the Radar
  • Authors