Brands Trust Us
Challenge: Speaker Recognition Algorithms
Performance of speaker ID algorithms in the presence of noises (we have only trained with some noises) and background voices. Deployment at scale is another issue.
Meeami Solution:
Speaker ID
DNN based (MFCC feature extraction and threshold-based classification)
Robust performance with scalable options for deployments with Siri, Alexa et all (SDK + Cloud based API)
In-built audio cleanup on mobile devices
Providing inbuilt real time noise removal, background voices removal, active noise cancellation, echo cancellation on Pixel and apple mobile devices.
Meeami Solution:
MVNS, BVS, AEC, ANC
MVNS – DNN based mask estimation, BVS – DNN based deep filter estimation, AEC – Soft attention for delay estimation and DNN based complex filtering
Real time multiple problem solving and robust performance.
In-built audio cleanup on landline/VoIP phones/intercom systems (GE)
Providing inbuilt real time noise removal, background voices removal, active noise cancellation, echo cancellation on wired phones
Meeami Solution:
MVNS, BVS, AEC, ANC
MVNS – DNN based mask estimation, BVS – DNN based deep filter estimation, AEC – Soft attention for delay estimation and DNN based complex filtering
Real time multiple problem solving and robust performance. Scalable options for deployment (SDK + Cloud based API).
See Meeami in Action
