Comparison of discriminative input and output transformations for speaker adaptation in the hybrid NN/HMM systems

By Bo Li, Khe Chai Sim. 2010

187 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, 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. Jinyu Li
2. Dong Yu
3. Li Deng
4. Florian Metze
5. Yifan Gong
6. Khe Chai Sim
7. James R. Glass
8. Yajie Miao

In Computer Vision and Pattern Recognition

1. Li Deng
2. Dong Yu
3. James R. Glass
4. Florian Metze
5. Wei-Ning Hsu
6. Kaisheng Yao
7. Yifan Gong
8. Jinyu Li

In Signal Processing

1. Jinyu Li
2. Dong Yu
3. Li Deng
4. Yifan Gong
5. Florian Metze
6. Khe Chai Sim
7. Yajie Miao
8. Chin-Hui Lee

Citations in the corpus

• Personalization of End-to-End Speech Recognition on Mobile Devices for Named Entities (2019, background): “There are existing speaker adaptation techniques for neural networks that offer a more compact model representation by having a large number of global parameters shared across different speakers and a smaller number of speaker-specific parameters [13, 14, 15, 16, 17, 18, 19, 20].”
• The CAPIO 2017 Conversational Speech Recognition System (2017, methodology): “As a result there have been alternative approaches to update only a small part of a neural network model [7, 8, 9, 10] to obtain adaptation benefits.”
• Modeling speaker variability using long short-term memory networks for speech recognition (2015, background): “Alternatively, a group of approaches perform adaptation by adding a linear transformation layer to the input [7], hidden [8], or output layer [9] of a DNN.”
• Hermitian Polynomial for Speaker Adaptation of Connectionist Speech Recognition Systems (2013, unknown): “, [40].”
• Maximum a posteriori adaptation of network parameters in deep models (2015, background): “The key idea is to augment the structure of the ANN component by adding an affine transformation network to the input [6], hidden [11], or output layer [22].”
• A Study of Enhancement, Augmentation, and Autoencoder Methods for Domain Adaptation in Distant Speech Recognition (2018, background): “Note that we do not consider the online adaptive setting, a common scenario for speaker adaptation [3, 4, 20, 21], where we have a small amount of labeled data to adapt to the target domain.”
• A Spectral Masking Approach to Noise-Robust Speech Recognition Using Deep Neural Networks (2014, methodology): “Linear Input Network (LIN) adaptation [29]–[33] represents training and testing feature mismatches with a weight matrix”
• Unsupervised Adaptation of Acoustic Models for ASR Using Utterance-Level Embeddings from Squeeze and Excitation Networks (2019, background): “In case of speaker adaptation, speaker-specific linear layers can be inserted either in input, output or in hidden layers [2],[3],[4].”
• Densely Connected Networks for Conversational Speech Recognition (2018, methodology): “As a result there have been alternative approaches to update only a small part of a neural network model [20, 21, 22, 23] to obtain adaptation benefits.”
• Speaker Adaptive Training of Deep Neural Network Acoustic Models Using I-Vectors (2015, methodology): “In [10], [2], [11], the SI-DNN models are augmented with additional speaker-dependent (SD) layers that are trained on the adaptation data.”
• Low-rank plus diagonal adaptation for deep neural networks (2016, methodology): “A simple and popular approach to combat the mismatch problem is to adapt the DNN model by applying a linear transformation on either input features, hidden layers, or the input to the softmax layer [5, 6, 7, 8, 9].”
• Fast Adaptation of Deep Neural Network Based on Discriminant Codes for Speech Recognition (2014, background): “Color versions of one or more of the .gures in this paper are available online at http://ieeexplore.ieee.org.”
• Adaptation of context-dependent deep neural networks for automatic speech recognition (2012, methodology): “The proposed feature-space discriminative linear regression or fDLR is similar to LIN [4, 7, 12] with two differences.”
• Embedding-Based Speaker Adaptive Training of Deep Neural Networks (2017, methodology): “Or, alternatively, the input or output layer is adapted by an affine transformation estimated using the adaptation data [6][7][8].”
• Transfer learning for speech and language processing (2015, background): “Some comparison for adaptation on different components can be found in [98], [99].”
• Stimulated Deep Neural Network for Speech Recognition (2016, methodology): “Several methods have been proposed to impose interpretable structures on the DNN topology to effectively adapt the DNN. Transformation-based schemes [8, 9, 10] add additional linear layers as speaker-dependent (SD) transforms.”
• An Investigation Into On-device Personalization of End-to-end Automatic Speech Recognition Models (2019, unknown): “A common solution to this problem is to adapt a well-trained model using a small amount of user-specific data to build a personalized model for each user [4, 5, 6, 7, 8, 9, 10, 11, 12, 13].”
• Cluster adaptive training for deep neural network (2015, background): “Linear transforms are also widely used to adapt feature or hidden-layer output [9, 10, 11, 12].”