Improving speaker recognition performance in the domain adaptation challenge using deep neural networks

By Daniel Garcia-Romero, Xiaohui Zhang, Alan McCree, Daniel Povey. 2014

97 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Daniel Povey
2. Sanjeev Khudanpur
3. Yoshua Bengio
4. Shinji Watanabe
5. James R. Glass
6. David Snyder
7. Daniel Garcia-Romero
8. Quoc V. Le

In Signal Processing

1. Daniel Povey
2. Sanjeev Khudanpur
3. David Snyder
4. Daniel Garcia-Romero
5. Tom Ko
6. Brian Kan-Wing Mak
7. Shinji Watanabe
8. James R. Glass

Citations in the corpus

• End-to-end text-dependent speaker verification (2015, background): “Hybrid approaches that include deep learning based components have also proved to be beneficial for text-independent speaker recognition [12, 13, 14].”
• Deep Neural Network Embeddings for Text-Independent Speaker Verification (2017, background): “Recent work has shown that incorporating an ASR DNN acoustic model can improve the UBM’s ability to model phonetic content [8, 9, 10, 11, 12].”
• Deep Neural Network Approaches to Speaker and Language Recognition (2015, methodology): “The DAC13 is a specified set of hyper-parameter , enroll, and test lists developed to exhibit a data domain shift for a SR task and has been reported on in several publications [17], [29], [30].”
• X-Vectors: Robust DNN Embeddings for Speaker Recognition (2018, unknown): “In either case, if the ASR DNN is trained on in-domain data, the improvement over traditional acoustic i-vectors is substantial [10, 11, 12].”
• A unified deep neural network for speaker and language recognition (2015, methodology): “While the DNN-posterior technique with MFCCs gives a significant gain over the baseline system for both sets of experiments, as also re-ported in [6]and [17], an even greater gain is realized us-ing bottleneck features with a GMM.”
• Deep neural network-based speaker embeddings for end-to-end speaker verification (2016, unknown): “Recently, this paradigm has been improved by incorporating deep neural network (DNN) acoustic models [8, 9, 10, 11, 12].”
• Self-Attentive Speaker Embeddings for Text-Independent Speaker Verification (2018, background): “Also, hybrid approaches that incorporate deep neural networks (DNNs) trained for automatic speech recognition (ASR) into the i-vector system have proved to be beneficial in some conditions [6, 7, 8, 9, 10].”
• Time delay deep neural network-based universal background models for speaker recognition (2015, methodology): “Usually, DNN-based speaker recognition systems employ a supervised-GMM derived from the DNN posteriors and speaker recognition features (often not the same as the DNN features) [8, 9, 10].”