Fast speaker adaptation of hybrid NN/HMM model for speech recognition based on discriminative learning of speaker code

By Ossama Abdel-Hamid, Hui Jiang. 2013

235 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

In Signal Processing

Citations in the corpus

Sample Efficient Adaptive Text-to-Speech (2018, background): “…studies have attempted to address the problem of few-shot learning by using deep neural networks, and they have shown promising results on classiﬁcation tasks in vision (Santoro et al., 2016; Shyam et al., 2017), language (Vinyals et al., 2016) and speech recognition (Abdel-Hamid and Jiang, 2013).”
Neural Voice Cloning with a Few Samples (2018, methodology): “The ﬁrst group is speaker adaptation for the whole-model [Yu et al., 2013], a portion of the model [Miao and Metze, 2015, Cui et al., 2017], or merely to a speaker embedding [Abdel-Hamid and Jiang, 2013, Xue et al., 2014].”
Deep Voice 2: Multi-Speaker Neural Text-to-Speech (2017, methodology): “The approach was investigated in speech recognition (Abdel-Hamid and Jiang, 2013), but is a novel technique in speech synthesis.”
Speaker adaptation of neural network acoustic models using i-vectors (2013, background): “Future work will address refining the i-vector inputs along the lines of the speaker code idea proposed in [4].”
Speaker Adaptive Training of Deep Neural Network Acoustic Models Using I-Vectors (2015, background): “append speaker codes [41], [42] to the hidden and output”
Domain Adaptation Using Factorized Hidden Layer for Robust Automatic Speech Recognition (2018, methodology): “Many DNN adaptation methods have been previously studied, including i-vector feature augmentation [19, 20], speaker-code adaptation [21], cluster adaptive training (CAT) [22, 23], factorized hidden layer [9], learning hidden unit contribution (LHUC) [24], singular value decomposition based subspace…”
Improving Robustness of Deep Neural Network Acoustic Models via Speech Separation and Joint Adaptive Training (2015, methodology): “In the context of DNN-AMs, a speaker adaptive training strategy was proposed in [1], where a discriminative speaker code and a non-linear feature transform are learned to transform the original features to the representation in the ﬁ rst hidden layer of the DNN-AM.”
On training the recurrent neural network encoder-decoder for large vocabulary end-to-end speech recognition (2016, background): “The past few years have seen signiﬁcant advancements in speech recognition based on this hybrid architecture including using different neural network architectures [4, 5, 6], sequence training [7, 8, 9] and speaker adaptation [10, 11, 12].”
Sentence-level control vectors for deep neural network speech synthesis (2015, methodology): “The idea is used for acoustic modelling in [14, 15, 16], where the CVs are speaker codes for rapidly adapting DNN-based speech recognition system to new speakers.”
Improving DNN speaker independence with I-vector inputs (2014, background): “[14] O. Abdel-Hamid and H. Jiang, “Rapid and effective speaker adaptation of convolutional neural network based models for speech recognition,” in Proc.”
Fast Adaptation of Deep Neural Network Based on Discriminant Codes for Speech Recognition (2014, methodology): “Finally, we have achieved very good performance in Switchboard (12.1% in WER) after speaker adaptation using sequence training criterion, which is very close to the best performance reported in this task ( Deep convolutional neural networks for LVCSR, T. N. Sainath et al., Proc.”
Speaker adaptation of context dependent deep neural networks based on MAP-adaptation and GMM-derived feature processing (2014, background): “• Non-linearfeature-spacetransformation, such as adaptation with speaker codes [17].”
Investigation of unsupervised adaptation of DNN acoustic models with filter bank input (2014, methodology): “Instead of modifying network parameters, several methods, such as linear input network (LIN) [11], feature discriminative linear regression (fDLR) [12], and the speaker code approach [13], discriminatively convert input features to improve DNN’s classification accuracy for the adaptation data.”
Maximum a posteriori adaptation of network parameters in deep models (2015, background): “Recent approaches, such as regularization based [7, 8], subspace based [9, 10], transformation based [6, 11, 12, 13], i-Vector based [14], native neural network based [15, 16, 17], factorization based [18] and fast adaptation schemes based on discriminant speaker codes [19, 16, 20, 21], have been pr”
Transfer learning for speech and language processing (2015, background): “An early study reported in [85] introduced a user vector (user code) to represent a speaker, and the vector was”
Speaker Representations for Speaker Adaptation in Multiple Speakers' BLSTM-RNN-Based Speech Synthesis (2016, methodology): “As described in [12], if there are K speakers’ corpora for model training, we can simply use zc = (z1,c, z2,c, · · · , zk,c, · · · zK,c) to represent the c-th speaker’s code.”
Multimodal Intelligence: Representation Learning, Information Fusion, and Applications (2019, unknown): “acoustic models [99], [100].”