Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore state-of-the-art methods for extracting speaker representations from speech in this comprehensive lecture delivered by Dr. Oldřich Plchot from Brno University of Technology at the JSALT 2025 workshop. Delve into current trends in speaker verification technology, comparing unsupervised, self-supervised, weakly supervised, and fully supervised approaches for generating speaker embeddings. Examine the rise of self-supervised Transformers and their integral role in modern speech modeling for automatic speech recognition, while discovering various applications and use cases that align with different methodological approaches. Learn from an expert researcher who has contributed significantly to international evaluations organized by NIST, IARPA, and DARPA, and has authored over 70 papers in prestigious venues including IEEE Transactions on Audio, Speech, and Language Processing, ICASSP, and Interspeech. Gain insights into the evolution of speaker verification systems and understand how emerging techniques are shaping the future of speech-based identity recognition technology.
