Researchers develop offline speech recognition that’s 97% accurate

Monday, October 22, 2018

Typically, deep learning approaches to voice recognition — systems that employ layers of neuron-mimicking mathematical functions to parse human speech — lean on powerful remote servers for bulk of processing. But researchers at the University of Waterloo and startup DarwinAI claim to have pioneered a strategy for designing speech recognition networks that not only achieves state-of-the-art accuracy, but which produces models robust enough to run on low-end smartphones.

Read the full story at VentureBeat

  1. 2018 (28)
    1. November (2)
    2. October (2)
    3. September (3)
    4. August (5)
    5. June (3)
    6. May (1)
    7. April (4)
    8. March (1)
    9. February (6)
    10. January (1)
  2. 2017 (25)
    1. December (2)
    2. November (4)
    3. October (2)
    4. September (4)
    5. August (3)
    6. July (1)
    7. June (3)
    8. May (2)
    9. March (3)
    10. January (1)
  3. 2016 (2)
    1. December (1)
    2. May (1)