-
Interactive Prototype Demo
Next-generation Lip-Reading Hearing Aids: Exploiting the power of contextual Big Data analytics
Amir Hussain, Ahsan Adeel, Mandar Gogate
KNOW MORE
Team
Professor Amir Hussain
Principal Investigator
Dr. Ahsan Adeel
Postdoctoral Scientist
Mandar Gogate
Doctoral Scientist
With acknowledgments to our collaborators: Prof. Roger Watt (University of Stirling), Dr. Jon Barker (University of Sheffield), Dr. Ricard Marxer (University of Sheffield), Ashraya Samba Shiva (University of Stirling), Dr. Andrew Abel (Xi'an Jiaotong-Liverpool University), William Whitmer (MRC/CSO Inst. Of Hearing Research), Dr. Peter Derleth (Sonova AG, Switzerland)
INTERACTIVE PROTOTYPE DEMO
Headphones are recommended for best experience (as hearing-aids are more or less
like headphones).
for Speech Enhancement
Spectrogram and Time Domain Signal Comparison
-
Benchmark vs. Proposed
Headphones are recommended for best experience (as hearing-aids are more or less like headphones).
The Man Called the Police
It’s Getting Cold in Here
They Ate the Lemon Pie
Proposed: Lip Reading Driven Deep Learning
Noisy (SNR ~ -10dB)
Recovered
Hello, My Name is Rachel
Noisy
Bin green A 9 again (SNR ~ -12dB)
Bin green A 9 again (SNR ~ -6dB)
Bin green A 9 again (SNR ~ 0dB)
Bin green A 9 again (SNR ~ 6dB)
Bin green A 9 again (SNR ~ 12dB)
SS
Bin green A 9 again (SNR ~ -12dB)
Bin green A 9 again (SNR ~ -6dB)
Bin green A 9 again (SNR ~ 0dB)
Bin green A 9 again (SNR ~ 6dB)
Bin green A 9 again (SNR ~ 12dB)
LMMSE
Bin green A 9 again (SNR ~ -12dB)
Bin green A 9 again (SNR ~ -6dB)
Bin green A 9 again (SNR ~ 0dB)
Bin green A 9 again (SNR ~ 6dB)
Bin green A 9 again (SNR ~ 12dB)
EVWF (proposed)
Bin green A 9 again (SNR ~ -12dB)
Bin green A 9 again (SNR ~ -6dB)
Bin green A 9 again (SNR ~ 0dB)
Bin green A 9 again (SNR ~ 6dB)
Bin green A 9 again (SNR ~ 12dB)