Comparison of Some Acoustic Noise Models and Their Effect on the Acoustic Diagnosis of Social Media Fingerprints

• The importance of preserving voiceprints as well as verifying their authenticity has increased, especially since reliance on them in the Corona period made many users rely on them in their work in directing administrative orders. As a result, this research came in an attempt to employ several algorithms for neural networks to verify voiceprints for (50-100 (1 person and for each person) (10-20) samples were taken. The results showed that the wavelet transform was affected by (the number of people, the number of sound signatures, and the noise). It was taken from one of the most famous social networking programs, which is (whatsApp), and the results showed that the design for each the type of noise and the type of filter adopted to reduce the effect of noise, reached the highest rating (99.2%) and it is due to the convergent neural network CNN (Band pass filter), while the worst rating reached 95.5% and it is due to the case of the convergent network CNN (AWGN) as shown Results The ability of some filters to increase the classification accuracy of the convention neural network and reduce the effect of noise.

Ämnesord

TEKNIK OCH TEKNOLOGIER  -- Maskinteknik -- Strömningsmekanik och akustik (hsv//swe)

ENGINEERING AND TECHNOLOGY  -- Mechanical Engineering -- Fluid Mechanics and Acoustics (hsv//eng)

Nyckelord

Computer Science

Datavetenskap

Publikations- och innehållstyp

ref (ämneskategori)

art (ämneskategori)