AsiaIPEX is a one-stop-shop for players in the IP industry, facilitating IP trade and connection to the IP world. Whether you are a patent owner interested in selling your IP, or a manufacturer looking to buy technologies to upgrade your operation, you will find the portal a useful resource.

Back to search results

Algorithm for real-time speech separation from a mixed sound signal

Summary: This technology is an algorithm called Deep Attractor Network (DANet) that isolates multiple sound sources from a single channel recording to enable individual speech recognition.

Technology Benefits: Automatic speech recognitionReal-time implementationInsensitive to speaker number or orderFaster processing and better performance than competitors

Technology Application: Speech recognition softwareHearing aids and other speech enhancement devicesNoise reduction in voice communications and noise-cancelling headphonesMusic enhancement (vocal track isolation)Security software or auditory forensics

Detailed Technology Description: None

*Abstract: None

*Inquiry: Beth KaudererColumbia Technology VenturesTel: (212) 854-8444Email: TechTransfer@columbia.edu

*IR: CU17276

*Principal Investigator

*Publications: Chenz Z, Luo Y, Mesgarani N. “Deep attractor network for single-microphone speaker separation” Proc IEEE Int Conf Acoust Speech and Signal Process. 2017 Mar 28.Tech Ventures Reference:IR CU17276Licensing Contact: Beth Kauderer

Country/Region: USA

For more information, please click Here