Spoken Language Processing: A Guide to Theory, Algorithm and System Development

By Xuedong Huang, Alex Acero, Hsiao-Wuen Hon, R. Reddy. 2001

573 citations recorded by the source corpus.

Subfields: Natural Language Processing, Speech & Audio

How researchers see this paper

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

In Natural Language Processing

In Speech & Audio

Citations in the corpus

A Diversity-Promoting Objective Function for Neural Conversation Models (2015, background): “Although the MMI optimization criterion has been comprehensively studied for other tasks, such as acoustic modeling in speech recognition (Huang et al., 2001), adapting MMI to SEQ2SEQ training is empirically nontrivial.”
Neural Approaches to Conversational AI (2018, neutral): “ve, unless such a bias emerges from the training data itself. While this argument may be true in general, optimizing the mutual information objective (also known as Maximum Mutual Information or MMI (Huang et al., 2001)) can be challenging, so Li et al. (2016a) used that objective at inference time.”
Reformulating the HMM as a trajectory model by imposing explicit relationships between static and dynamic feature vector sequences (2004, methodology): “However, some limitations for modeling sequence of speech spectra by the standard HMM have been pointed out (Ostendorf et al., 1996; Nakagawa, 2002; Huang et al., 2001).”
The remarkable robustness of surrogate gradient learning for instilling complex function in spiking neural networks (2020, methodology): “From the power spectrum we further computed the filter banks by applying 40 triangular filters on a Mel-scale [51].”
Automatic music transcription: challenges and future directions (2013, background): “specific, and many modern speech recognisers include speaker adaptation [65].”
Deep learning for minimum mean-square error approaches to speech enhancement (2019, methodology): “Signal processing, noise estimation, and a posteriori SNR estimation The Hamming window function was used for analysis and synthesis Picone, 1993; Huang et al., 2001; Paliwal and Wojcicki, 2008 ), ith a frame length of 32 ms ( 𝑁 𝑙 = 512 ) and a frame shift of 16 ms 𝑁 𝑠 = 256 ).”
An overview of text-independent speaker recognition: From features to supervectors (2010, methodology): “The coefficients ak are usually determined by minimizing the residual energy using the so-called Levinson-Durbin algorithm [82, 102, 189].”
A fast maximum likelihood nonlinear feature transformation method for GMM-HMM speaker adaptation (2014, background): “A comprehensive review of existing techniques can be found in [3].”