Parallel Networks that Learn to Pronounce English Text

By Terrence J. Sejnowski, Charles R. Rosenberg. 1987

How researchers see this paper

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

Citations in the corpus

An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling (2018, methodology): “Convolutional networks (LeCun et al., 1989) have been applied to sequences for decades (Sejnowski & Rosenberg, 1987; Hinton, 1989).”
Rethinking eliminative connectionism. (1998, methodology): “For example, one suggested that the identity task could be solved by a model that used a ‘‘movingwindow’’ technique borrowed from Rosenberg and Sejnowski (1987), along the lines of the model illustrated in Fig.”
A Simple Weight Decay Can Improve Generalization (1991, methodology): “The error rate is the fraction of wrong phonemes when the phoneme vector with the smallest angle to the actual output is chosen, see [11].”
Robust Anomaly Detection for Multivariate Time Series through Stochastic Recurrent Neural Network (2019, methodology): “During data preprocessing, the dataset is transformed by data standardization, and then it is segmented into sequences through sliding windows [21] of length T + 1.”
Stochastic Learning (2003, background): “Readers interested by the practice of stochastic gradient algorithms should also read [25] and investigate applied contributions such as [39, 37, 46, 6, 24, 26].”
Fast neural net simulation with a DSP processor array (1995, background): “The latter network has the dimensions of Nettalk [19], a neural netthat learns to convert written English text to speech.”
Benchmarking Least Squares Support Vector Machine Classifiers (2004, methodology): “A straightforward extension of LS-SVMs to multiclass problems has been proposed in Suykens and Vandewalle (1999c), where additional outputs are taken in order to encode multiclasses as is often done in classical neural networks methodology (Bishop, 1995).”
Combining Symbolic and Neural Learning (1994, unknown): “For example, Sejnowski and Rosenberg (1987) manually analyzed clusters developed on the NETtalk task, with some success.”
Text-to-Speech Synthesis (2009, methodology): “The first data driven G2P algorithm was the NetTalk algorithm developed by Sejnowski and Rosenberg [397], [396], [377].”
Merging chrominance and luminance in early, medium, and late fusion using Convolutional Neural Networks (2015, background): “In the same year, Sejnowski and Rosenberg demonstrated the usefulness of MFNN with NETtalk [Sejnowski and Rosenberg, 1987], a neural network that learned to pronounce English text.”
30 years of adaptive neural networks: perceptron, Madaline, and backpropagation (1990, background): “These procedures are the main focus of this paper.”
Recursive Distributed Representations (1990, background): “One way to overcome this length limitation is by "sliding" the input across the buffer [15, 16].”
On Learning Vector − Valued Functions (2003, methodology): “This problem was studied by Sejnowski and Rosenberg in their NETtalk system [25].”
Integration of New Information in Memory: New Insights from a Complementary Learning Systems Perspective (2020, unknown): “Several models of this kind were presented in the late 1980’s and early 1990’s, demonstrating how the mastery of structured bodies of knowledge of several different types could occur in this way, including knowledge of the mapping from spelling to sound [1, 2], knowledge of the syntactic structure o”