Following this success, the team embarked on research to develop neural networks for advanced telephone conversation analysis. This research led to the creation of several products, including a proprietary speech-to-text (STT) system and various models for automatic conversation analysis, enhancing the ability to monitor and improve customer service quality.
In 2021, the team developed the ERANN neural network algorithm, which achieved global recognition for its superior accuracy in sound event recognition.
The project team was established in 2020 by specialists in neural network technologies and artificial intelligence in Novosibirsk Akademgorodok.
Team history
Crafting Solutions Where Others Struggle
An enhanced neural network model for speech-to-text (STT) conversion in phone calls has been developed.
May 2024
A neural network system has been developed to monitor horse health using advanced CV techniques.
December 2023
A neural network model has been developed to detect negativity in the human voice.
December 2023
A neural network model has been developed to detect fatigue in an operator's voice.
October 2023
A GPT model has been developed for summarizing phone calls.
August 2023
A neural network model has been developed to identify potential leads from phone calls.
July 2023
A neural network model has been developed to detect extraneous noise in phone calls.
May 2023
A neural network has been developed for the automatic recognition of synthesized speech in telephone conversations.
October 2022
The DARWIN AI platform has been developed for automatic analysis and quality control of phone calls.
August 2022
The team participated in the international DCASE2022 competition, achieving 21st place out of 83 teams.
July 2022
June 2022
A model has been developed to automatically insert punctuation marks into recognized text.
May 2022
A high-speed model for detecting human voice has been developed.
AI Algorithms Development
A model has been developed for speech-to-text (STT) conversion designed for low-quality telephone calls.
April 2022
An algorithm has been developed for diagnosing COPD in horses using auscultation sounds of the respiratory system.
September 2021
The team competed in the international DCASE2021 competition, securing 7th place out of 31 teams.
July 2021
An algorithm for diagnosing human respiratory conditions through sound analysis has been developed, drawing interest from the company AstraZeneca.
June 2021
Developed the ERANN neural network architecture, which achieved global recognition as the most accurate for classifying sound events in 2021.
May 2021
The team participated in international Kaggle competition, achieving a ranking in the Top 3% with an 35th place out of 1,143 teams.
April 2021
An algorithm has been developed for detecting and classifying the causes of a child's crying.
March 2021
A neural network model has been developed for analyzing pauses in phone calls.
November 2023
A neural network model has been developed to identify positive and negative interruptions during conversations.
September 2023
