Structured speech modeling

By Li Deng, Dong Yu, A. Acero. 2006

100 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. Li Deng
2. Dong Yu
3. Jinyu Li
4. Yoshua Bengio
5. Tara N. Sainath
6. Andrew W. Senior
7. Quoc V. Le
8. Jianfeng Gao

In Signal Processing

1. Li Deng
2. Dong Yu
3. Jinyu Li
4. Tara N. Sainath
5. Andrew W. Senior
6. Yifan Gong
7. M. Seltzer
8. Yoshua Bengio

Citations in the corpus

• Deep Learning: Methods and Applications (2014, methodology): “Further, the DNN results were found to be slightly superior to the then-best-performing single system based on the generative hidden trajectory model (HTM) in the literature [105, 108] evaluated on the same, commonly used TIMIT task by many speech researchers [107, 108, 274, 313].”
• A tutorial survey of architectures, algorithms, and applications for deep learning (2014, background): “Generative models with deep hidden dynamics likewise have not been competitive either (Picone et al., 1999; Deng et al., 2006).”
• An Overview of Deep-Structured Learning for Information Processing (2011, background): “After learning, the values of the latent variables in every layer can be inferred by a single, bottom-up pass that starts with an observed data vector in the bottom layer and uses the generative weights in the reverse direction.”
• Statistical approach to vocal tract transfer function estimation based on factor analyzed trajectory HMM (2008, methodology): “This framework is similar to the structured speech modeling [7] in terms of using a state space model for modeling an observation sequence.”
• Deep Belief Networks for phone recognition (2009, background): “In [11], a probabilistic generative model is introduced where the dynamic structure in the hidden vocal tract resonance space is used to characterize long-span contextual influence across phonetic units.”
• Use of Differential Cepstra as Acoustic Features in Hidden Trajectory Modeling for Phonetic Recognition (2007, methodology): “They are determined by the VTR-targets' covariance parameters of the HTM and by the coarticulation properties of the VTR dynamics as elaborated in [4], and are considered fixed for the discussion ofthe cepstral prediction residual parameters as the focus of this paper.”
• A Novel Framework and Training Algorithm for Variable-Parameter Hidden Markov Models (2009, background): “Many efforts have been put during the past two decades in search for new modeling, adaptation, and training technologies (e.g., [1], [3], [5], [6], [9], [10], [13], [14], [21], [24], [25], [27]–[29]) that can deliver sufficiently high recognition accuracy under all deployment conditions.”
• Acoustic Modeling Using Deep Belief Networks (2012, unknown): “For each featur , a specialized acoustic representation is constructed in whch that feature is easy to detect.”
• Deep Convex Net: A Scalable Architecture for Speech Pattern Classification (2011, methodology): “(We are nevertheless clearly aware that the unique elasticity of temporal dynamic of speech as explained in [1] would require temporally-correlated models better than HMM for the ultimate success of ASR, and integrating such a model with the DCN to form the coherent dynamic DCN is by itself a more…”
• A deep convolutional neural network using heterogeneous pooling for trading acoustic invariance with phonetic confusion (2013, methodology): “An example of such speech variability is formant undershooting (or overshooting) caused by casual (or forced) speaking styles that have been mathematically represented by hidden dynamic or trajectory models of speech [11][12][13] [14].”
• Investigation of full-sequence training of deep belief networks for speech recognition (2010, methodology): “After decoding, the 61 phone classes were mapped to a standard set of 39 classes for scoring using the HResults tool in HTK, in the same way as other standard TIMIT experiments (e.g., [5]).”
• Deep Neural Networks for Acoustic Modeling in Speech Recognition (2012, methodology): “This is because temporal variations in speech can be partially handled by the dynamic programing procedure in the HMM component and those aspects of temporal variation that cannot be adequately handled by the HMM can be addressed more explicitly and effectively by hidden trajectory models [40].”
• Statistical Parametric Speech Synthesis (2007, methodology): “One possible extension of Ling et al.’s technique is using structured speech models, which can include hidden levels of speech production (Richards and Bri-dle, 1999; Rosti and Gales, 2003; Deng et al., 2006; Frankel and King, 2007; Frankel et al., 2007).”
• New types of deep neural network learning for speech recognition and related applications: an overview (2013, background): “[14][15], in the literature and evaluated on the identical task.”
• Binary coding of speech spectrograms using a deep auto-encoder (2010, background): “…of generative modeling and of exploiting the multiple layers of non-linear transformations in the human speech generation process (e.g., [1][2][3][4][5][6]) it was not until the introduction of generative, layer-by-layer pre-training that the full power of generative modeling became apparent.”
• Tensor Deep Stacking Networks (2013, unknown): “2 % GMM-HMM [24] 28.”