MTGAN: Speaker Verification through Multitasking Triplet Generative Adversarial Networks
arXiv:1803.09059
Abstract
In this paper, we propose an enhanced triplet method that improves the encoding process of embeddings by jointly utilizing generative adversarial mechanism and multitasking optimization. We extend our triplet encoder with Generative Adversarial Networks (GANs) and softmax loss function. GAN is introduced for increasing the generality and diversity of samples, while softmax is for reinforcing features about speakers. For simplification, we term our method Multitasking Triplet Generative Adversarial Networks (MTGAN). Experiment on short utterances demonstrates that MTGAN reduces the verification equal error rate (EER) by 67% (relatively) and 32% (relatively) over conventional i-vector method and state-of-the-art triplet loss method respectively. This effectively indicates that MTGAN outperforms triplet methods in the aspect of expressing the high-level feature of speaker information.
arXiv commentsubmitted to Interspeech 2018
References in corpus (7)
- Conditional Generative Adversarial Nets
- In Defense of the Triplet Loss for Person Re-Identification
- Improved Training of Wasserstein GANs
- Deep Speaker: an End-to-End Neural Speaker Embedding System
- Conditional Generative Adversarial Networks for Speech Enhancement and Noise-Robust Speaker Verification
- Sampling Matters in Deep Embedding Learning
- TripletGAN: Training Generative Model with Triplet Loss