DeepRead
Cover of Statistical methods for speech recognition

Statistical methods for speech recognition

Frederick Jelinek · 1997

Machine learning, statistics and probability › Bayesian and probabilistic modeling

#13,268 most recommended

About the book

The book covers the statistical foundations of automatic speech recognition, treating it as the problem of finding the most probable word sequence given an acoustic signal. It spans the mathematics of information theory, language modeling, acoustic modeling with hidden Markov models, and the search algorithms used for decoding. Jelinek's central claim is that speech recognition is best framed as a source-channel problem, where a language model supplies prior probabilities over word sequences and an acoustic model supplies the likelihood of the signal given those words. He argues that parameters for both models should be estimated from data rather than hand-crafted from linguistic rules, using methods such as maximum likelihood and the EM algorithm. The book also argues that efficient decoding, through techniques like Viterbi and stack search, makes this statistical approach practical at scale.

Written by Claude from what it knows of this book.

What Hacker News says 100% recommend · 0% critical

012201020122014201620182010 peak · 1
This bookAverage book in Bayesian and probabilistic modeling

In the commenters’ own words

What Open Library readers say

No Open Library reader has rated it yet.

  • 2 want to read
  • 1 reading
  • 0 have read

Open Library’s page for the book

What critics say, via Wikipedia

No Wikipedia article about this book yet, so no critics to quote.

Reddit communities — coming later

Details

First published
1997
ISBN
9780262100663
Editions known
2 at Open Library
Open Library
OL2714036W

Automatic speech recognitionComputational linguisticsMathematical linguisticsStatistical methods

Explore in the mosaic

Read next: its neighbours on the mosaic