Golden Gemini is All You Need: Finding the Sweet Spots for Speaker Verification
Previous studies demonstrate the impressive performance of residual neural networks (ResNet) in speaker verification. The ResNet models treat the time and frequency dimensions equally. They follow the default stride configuration designed for image recognition, where the horizontal and vertical axes exhibit similarities. This approach ignores the fact that time and frequency are asymmetric in speech representation. In this paper, we address this issue and look for optimal stride configurations specifically tailored for speaker verification. We represent the stride space on a trellis diagram, and conduct a systematic study on the impact of temporal and frequency resolutions on the performance and further identify two optimal points, namely Golden Gemini, which serves as a guiding principle for designing 2D ResNet-based speaker verification models. By following the principle, a state-of-the-art ResNet baseline model gains a significant performance improvement on VoxCeleb, SITW, and CNCeleb datasets with 7.70%/11.76% average EER/minDCF reductions, respectively, across different network depths (ResNet18, 34, 50, and 101), while reducing the number of parameters by 16.5% and FLOPs by 4.1%. We refer to it as Gemini ResNet. Further investigation reveals the efficacy of the proposed Golden Gemini operating points across various training conditions and architectures. Furthermore, we present a new benchmark, namely the Gemini DF-ResNet, using a cutting-edge model.
Code (1)
Tasks
AllSpeaker VerificationMethods 이 논문이 사용한 방법론
Similar Papers 제목 키워드 기반
Managing Data Lineage of O&G Machine Learning Models: The Sweet Spot for Shale Use Case
Machine Learning (ML) has increased its role, becoming essential in several industries. However, questions around training data lineage, such as "where has the dataset used to train this model come from?"; the introducti…
BIG-bench Machine LearningReproducing the Acoustic Velocity Vectors in a Circular Listening Area
Acoustic velocity vectors are important for human's localization of sound at low frequencies. This paper proposes a sound field reproduction algorithm, which matches the acoustic velocity vectors in a circular listening …
Reproducing the Acoustic Velocity Vectors in a Spherical Listening Region
Acoustic velocity vectors (AVVs) are related to the human's perception of sound at low frequencies and are widely used in Ambisonics. This paper proposes a spatial sound field reproduction algorithm called velocity match…
Optimizing transformer-based machine translation model for single GPU training: a hyperparameter ablation study
In machine translation tasks, the relationship between model complexity and performance is often presumed to be linear, driving an increase in the number of parameters and consequent demands for computational resources l…
GPUMachine TranslationTranslationWhen Does Pruning Benefit Vision Representations?
Pruning is widely used to reduce the complexity of deep learning models, but its effects on interpretability and representation learning remain poorly understood. This paper investigates how pruning influences vision mod…
Object DiscoveryRepresentation Learning