Probabilistic Spherical Discriminant Analysis: An Alternative to PLDA for length-normalized embeddings

Probabilistic Spherical Discriminant Analysis: An Alternative to PLDA for length-normalized embeddings

Stať ve sborníku v databázi WoS či Scopus

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.

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.

speaker recognition, PSDA, Von Mises-Fisher

speaker recognition, PSDA, Von Mises-Fisher

BRUMMER, J.; SWART, A.; MOŠNER, L.; SILNOVA, A.; PLCHOT, O.; STAFYLAKIS, T.; BURGET, L.

2023

18.09.2022

International Speech Communication Association

Incheon

Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH

1990-9772

Proceedings of Interspeech

2022

9

Francouzská republika

1446

1450

5

https://www.isca-speech.org/archive/pdfs/interspeech_2022/brummer22_interspeech.pdf

http://hdl.handle.net/

brummer_interspeech2022_probabilistic