Tuesday, June 27, 2017

Five-star reviews

You can tell who my friends are:  check out the nice 5-star reviews on Amazon

Reviewer Jordan Cohen is referenced in the book as probably the first to show a big improvement in speech recognition performance by moving to a more auditory front-end, at IBM in the 80s (after he turned down my job offer that he alludes to in his review). 

Reviewer Lloyd Watts did a lot of great work on cochlear modeling, as a grad student at Caltech, and later founded Audience to apply those ideas to valuable machine hearing problems such as noise suppression in mobile phones; I reference a number of his results, and use one of his figures.  Thanks, guys!

Of course, big thanks also to friends Roy Patterson, Bill Yost, and Dan Levitin, who wrote the "advance praise" blurbs on the back cover, and who have encouraged this line of work for decades.

More reviews, wherever you got the book, are good for me, for the field, and for the market.  Thanks, readers!

Monday, June 19, 2017


  • First bug report, by Erick Gallun: in first sentence of section 6.2, "to an output function to time" should be "to an output function of time." He gets the first bug bounty dollar, and my thanks (note that I edited this post after he commented below; hope that's not confusing).
  • Jeff Rector earns a dollar for spotting "simultaeous", which should be "simultaneous", in section 4.5, page 55. 
  • Tomek Maka earns a buck for the misformatted equation (paren in wrong place) in the box "Example: Delay Lines and Moving-Average Filters" in chapter 12, page 220.  Should be H(s)=\frac{Y(s)}{X(s)}=\exp(-sT). 
  • Tomek Maka finds one more stray paren:  "h(t))" has an extra close paren, middle equation on page 123.
  • Responding to my challenge to readers on section 17.2, AndrĂ© van Schaik reported the typo on page 311, second paragraph, where "a sampling term" should be "a damping term".
  • Two bugs from Wan-Teh's comment below: 
    • P.128, Sec. 7.5: In the second-to-last equation on this page: H(z)X(z) = Z{h(k) * x(k)} should have square brackets for indices, as H(z)X(z) = Z{h[k] * x[k]}.  
    • P. 132, Sec. 7.7: In the box "Detail: Zeros at the Origin", the last paragraph says: "s-domain points that would map to z = 0 are infinitely far to the right side in the s plane" – "right" should be changed to "left".
  • Roy Patterson reports a flubbed wording in the last paragraph of Chapter 20, p. 354: "continue to get away with ignore" should be "continue to ignore", or possibly "continue to get away with ignoring".  I'll add a dollar to the envelope I'm taking to give him in Cambridge.
  • A few errors and nits that I've found myself, that I'll fix in a later printing if I can.
    • Section 2.2 p. 26 "introduced below as" should really refer to a section: "introduced in section 2.5 as".
    • Section 2.3 p. 27 "as "Cooper et al. (2008) summarizes", should use the plural verb "summarize" (because in my style, I interpret the subject here as the authors, plural, not the paper, singular).
    • Figure 3.4. The dotted lines connecting the parts of the figure are too light to see.
    • Before the lower box on p. 224, "we studies" should be "we studied".
    • In the Physics Connection box on p. 228, "nonmoving sinusoidal spatial envelopes" should be singular.
    • In the box "On Quasi-Linear Filters" on p.240, "auditory filters models" should be "auditory filter models".
    • Figure 16.6 caption refers to "the middle damping" where it should say "the lowest damping"; or I could change the figure to match the caption.
    • Section 18.3, p. 328, "input to the ICS." should be "input to the IHCs."
    • In the box "History Connection: From IPD to ITD", p. 382, Figure 22.1 and caption have been placed badly, between the sentence introducing a quote and the quote.
    • Figure 25.1 caption, p. 443, "possbiility" should be "possibility".
I expect we'll find more.  Respond here or email me.

Here are bugs found in 2018, after the corrected PDF was out:

  • Binary logarithmss has an extra "s"; Chapter 3, box The Mathematics of Logarithms and Power Laws, second paragraph (p.34 in hardback); reported by Ben Caswell-Midwinter
  • Hershel is a misspelling of Herschel, and the Herschel 1930 ref should be 1830, in section 4.8 (p.67 in hardback).

Friday, June 16, 2017

Available in "UK and Europe"

Human and Machine Hearing is now launching in Europe.  Publisher says "Cambridge University Press is delighted to confirm that Human and Machine Hearing will publish on 15 June 2017 in the UK and Europe."

  • Book Depository has a great price (£51.05, €58.38, $65.37), all currencies, and free shipping worldwide.
  • Amazon.co.uk has it for £48.99 (that's 11% off the £54.99 list price).  June 17 update: the discount appears to have gone away; it's now listed at full price; keep an eye on it.
  • Blackwell's has it at £48.99 with free UK delivery.
  • Amazon.de has a modest discount, to match buch24.de and buecher.de at EUR 66.99; Amazon.fr looks expensive.
  • Cambridge University Press is offering a 20% discount at their UK site; use discount code LYON2017 at checkout; this work for pounds and dollars – haven't tried Euros.
  • I don't know what other book sellers are big in Europe; let me know.