Publication Details
Probabilistic Spherical Discriminant Analysis: An Alternative to PLDA for length-normalized embeddings
Swart Albert du Preez
Mošner Ladislav, Ing. (DCGM)
Silnova Anna, M.Sc., Ph.D. (DCGM)
Plchot Oldřich, Ing., Ph.D. (DCGM)
Stafylakis Themos
Burget Lukáš, doc. Ing., Ph.D. (DCGM)
speaker recognition, PSDA, Von Mises-Fisher
In speaker recognition, where speech segments are mapped to embeddings on the unit hypersphere, two scoring backends are commonly used, namely cosine scoring or PLDA. Both have advantages and disadvantages, depending on the context. Cosine scoring follows naturally from the spherical geometry, but for PLDA the blessing is mixedlength normalization Gaussianizes the between-speaker distribution, but violates the assumption of a speaker-independent within-speaker distribution. We propose PSDA, an analogue to PLDA that uses Von Mises- Fisher distributions on the hypersphere for both within and between-class distributions. We show how the self-conjugacy of this distribution gives closed-form likelihood-ratio scores, making it a drop-in replacement for PLDA at scoring time. All kinds of trials can be scored, including single-enroll and multienroll verification, as well as more complex likelihood-ratios that could be used in clustering and diarization. Learning is done via an EM-algorithm with closed-form updates. We explain the model and present some first experiments.
@inproceedings{BUT179687,
author="Johan Nikolaas Langenhoven {Brummer} and Albert du Preez {Swart} and Ladislav {Mošner} and Anna {Silnova} and Oldřich {Plchot} and Themos {Stafylakis} and Lukáš {Burget}",
title="Probabilistic Spherical Discriminant Analysis: An Alternative to PLDA for length-normalized embeddings",
booktitle="Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH",
year="2022",
journal="Proceedings of Interspeech",
volume="2022",
number="9",
pages="1446--1450",
publisher="International Speech Communication Association",
address="Incheon",
doi="10.21437/Interspeech.2022-731",
issn="1990-9772",
url="https://www.isca-speech.org/archive/pdfs/interspeech_2022/brummer22_interspeech.pdf"
}