I keep searching for the perfect gadget to take
my library to the beach, but haven't found it yet.
For the manufacturers out there, here would be
the holy grail, in opinion, for people who want to read:

(1) Around a 6" x 9" display (this is a fairly standard trade paperback
size).

(2) weighs a pound or less

(3) no keyboard, a few buttons, and a good voice recognition-supported
operating system
The purpose of the operating is *mainly* to retrieve (from the net or
my local pc) and display
.pdfs and .html content in a way as pleasing as a good paperback.
Anything else is nice, but icing on the cake.

(4) short and long-range connectivity for the retrieval in (3).

(5) a bright display for outdoors, in a rugged case that closes, and keeps
out the weather and sand :-)

In other words, its *like a book* with replaceable content in all important
aspects.

I know there have been some ebook-type reader marketing failures, but I
think it's
because no one has 'got it right' yet. I would be interested in hearing what
other
readers would consider 'getting it right', as I continue my search.

regards,
alan

On Sat, 28 May 2005 12:42:29 -0700, "Alan"
<info@optioncity.REMOVETHIS.net> wrote:

>I know there have been some ebook-type reader marketing failures, but I
>think it's because no one has 'got it right' yet. I would be interested in
>hearing what other readers would consider 'getting it right', as I
>continue my search.

I'll reserve judgment on whether it's possible to "get it right,"
since up until last week I never in my life seriously considered
buying an e-book. (I am buying one soon because a particular book I
want to read is available only in that format. I will probably print
the pages out to read them.)

That said, I think that if the e-book concept is viable at all, the
limitations of current display technology are the main thing holding
it back. LCD displays are too low in resolution to be read easily for
long periods of time, and too reflective for use in brightly
illuminated places. The lack of bandwidth (displaying only one page at
a time) is also a major drawback, a practical one for some users and a
psychological one for others. I think that when and if the e-book
becomes popular, it will consist of a bound volume of electronic
paper, rather than a "slab" with an LCD screen.

I will caution you about the use of voice recognition.
Limited-vocabulary voice recognition can be pretty reliable, but its
performance is likely to be seriously degraded by background noise.
Until their attention is drawn to it, many people do not realize how
loud the background noise is in many common environments: buses, city
streets, even the beach. I would be very skeptical of an e-book reader
whose user interface depended on voice recognition, and considering
the resources required, I would question the wisdom of offering it as
an option.

My email address is LLM041103 at earthlink dot net.

"Jonathan Sachs" <xxxxxxx@earthlink.not> wrote in message
news:thmh919jtg4r571utpdnvfvma1ssf4sr7e@4ax.com...
> On Sat, 28 May 2005 12:42:29 -0700, "Alan"
> <info@optioncity.REMOVETHIS.net> wrote:

> That said, I think that if the e-book concept is viable at all, the
> limitations of current display technology are the main thing holding
> it back. LCD displays are too low in resolution to be read easily for
> long periods of time, and too reflective for use in brightly
> illuminated places. The lack of bandwidth (displaying only one page at
> a time) is also a major drawback, a practical one for some users and a
> psychological one for others. I think that when and if the e-book
> becomes popular, it will consist of a bound volume of electronic
> paper, rather than a "slab" with an LCD screen.

Thanks for your feedback, Jonathan.
Since I want my e-book to have a cover, if you opened that and had
two 6 x 9 pages displayed, that would be even better for me.
However, the need to have multiple pages that you can physically
'turn' seems to me entirely psychological unless I am missing something.
I don't feel that need at all.

>
> I will caution you about the use of voice recognition.
> Limited-vocabulary voice recognition can be pretty reliable, but its
> performance is likely to be seriously degraded by background noise.
> Until their attention is drawn to it, many people do not realize how
> loud the background noise is in many common environments: buses, city
> streets, even the beach. I would be very skeptical of an e-book reader
> whose user interface depended on voice recognition, and considering
> the resources required, I would question the wisdom of offering it as
> an option.

You could be right --- it's perhaps the least necessary of my 'specs', and I
would
likely jump at a reasonably priced device with all my other 'requirements'
but not that one
My tv remote shows me that you can get pretty far with a few buttons.
Yet ... you need to be able to do a search on author, title, keywords,
etc.
I wonder if a requirement that you talk "into the open e-book",
combined with a few strategically placed pickup microphones, couldn't
substantially filter out the background? Or, how about a wireless, clip-on
microphone? Of course, at the beach, I'll have to clip it onto my hat :-)
Surely the acoustic engineers can solve this little problem.

regards,
alan

I dont think you're really taking into consideration background noise...
considering the waves themselves generate great amounts of "White Noise" (no
reference to the movie)... the only way of actually getting a clear voice in
a situation like that is to either do noise cancellation, or remove the back
ground all together... "More microphones" isnt going to help the situation.
What would need to be done for noise cancellation is record background noise
(when no one is talking) generate a 'noise map'... and when it recognizes
speech, or something it considers speech, apply the noise map in the noise
reduction... and voilah! clear voice, or nothing at all!... but this is very
cpu intensive... but can be done, and i've done it for years with now
outdated software, "cool edit 2000" great stuff.
-James G.

"Alan" <info@optioncity.REMOVETHIS.net> wrote in message
news:ejL%23Lz9YFHA.2128@TK2MSFTNGP15.phx.gbl...
>
> "Jonathan Sachs" <xxxxxxx@earthlink.not> wrote in message
> news:thmh919jtg4r571utpdnvfvma1ssf4sr7e@4ax.com...
>> On Sat, 28 May 2005 12:42:29 -0700, "Alan"
>> <info@optioncity.REMOVETHIS.net> wrote:
>
>> That said, I think that if the e-book concept is viable at all, the
>> limitations of current display technology are the main thing holding
>> it back. LCD displays are too low in resolution to be read easily for
>> long periods of time, and too reflective for use in brightly
>> illuminated places. The lack of bandwidth (displaying only one page at
>> a time) is also a major drawback, a practical one for some users and a
>> psychological one for others. I think that when and if the e-book
>> becomes popular, it will consist of a bound volume of electronic
>> paper, rather than a "slab" with an LCD screen.
>
> Thanks for your feedback, Jonathan.
> Since I want my e-book to have a cover, if you opened that and had
> two 6 x 9 pages displayed, that would be even better for me.
> However, the need to have multiple pages that you can physically
> 'turn' seems to me entirely psychological unless I am missing something.
> I don't feel that need at all.
>
>>
>> I will caution you about the use of voice recognition.
>> Limited-vocabulary voice recognition can be pretty reliable, but its
>> performance is likely to be seriously degraded by background noise.
>> Until their attention is drawn to it, many people do not realize how
>> loud the background noise is in many common environments: buses, city
>> streets, even the beach. I would be very skeptical of an e-book reader
>> whose user interface depended on voice recognition, and considering
>> the resources required, I would question the wisdom of offering it as
>> an option.
>
> You could be right --- it's perhaps the least necessary of my 'specs', and
> I would
> likely jump at a reasonably priced device with all my other 'requirements'
> but not that one
> My tv remote shows me that you can get pretty far with a few buttons.
> Yet ... you need to be able to do a search on author, title, keywords,
> etc.
> I wonder if a requirement that you talk "into the open e-book",
> combined with a few strategically placed pickup microphones, couldn't
> substantially filter out the background? Or, how about a wireless, clip-on
> microphone? Of course, at the beach, I'll have to clip it onto my hat :-)
> Surely the acoustic engineers can solve this little problem.
>
> regards,
> alan
>
>
>
>

well, I take that back, an extra microphone may be usefull... as the machine
notices talking from the user, it can also use another mic that is facing
away from the user. thus recording background noise... and then it can be
applied... but still cpu intensive. =)

-James G.


"James Gockel" <flibbertigibbet007_at_hotmail_dot_com> wrote in message
news:unnCNO%23YFHA.3620@TK2MSFTNGP09.phx.gbl...
>I dont think you're really taking into consideration background noise...
>considering the waves themselves generate great amounts of "White Noise"
>(no reference to the movie)... the only way of actually getting a clear
>voice in a situation like that is to either do noise cancellation, or
>remove the back ground all together... "More microphones" isnt going to
>help the situation.
> What would need to be done for noise cancellation is record background
> noise (when no one is talking) generate a 'noise map'... and when it
> recognizes speech, or something it considers speech, apply the noise map
> in the noise reduction... and voilah! clear voice, or nothing at all!...
> but this is very cpu intensive... but can be done, and i've done it for
> years with now outdated software, "cool edit 2000" great stuff.
> -James G.
>
> "Alan" <info@optioncity.REMOVETHIS.net> wrote in message
> news:ejL%23Lz9YFHA.2128@TK2MSFTNGP15.phx.gbl...
>>
>> "Jonathan Sachs" <xxxxxxx@earthlink.not> wrote in message
>> news:thmh919jtg4r571utpdnvfvma1ssf4sr7e@4ax.com...
>>> On Sat, 28 May 2005 12:42:29 -0700, "Alan"
>>> <info@optioncity.REMOVETHIS.net> wrote:
>>
>>> That said, I think that if the e-book concept is viable at all, the
>>> limitations of current display technology are the main thing holding
>>> it back. LCD displays are too low in resolution to be read easily for
>>> long periods of time, and too reflective for use in brightly
>>> illuminated places. The lack of bandwidth (displaying only one page at
>>> a time) is also a major drawback, a practical one for some users and a
>>> psychological one for others. I think that when and if the e-book
>>> becomes popular, it will consist of a bound volume of electronic
>>> paper, rather than a "slab" with an LCD screen.
>>
>> Thanks for your feedback, Jonathan.
>> Since I want my e-book to have a cover, if you opened that and had
>> two 6 x 9 pages displayed, that would be even better for me.
>> However, the need to have multiple pages that you can physically
>> 'turn' seems to me entirely psychological unless I am missing something.
>> I don't feel that need at all.
>>
>>>
>>> I will caution you about the use of voice recognition.
>>> Limited-vocabulary voice recognition can be pretty reliable, but its
>>> performance is likely to be seriously degraded by background noise.
>>> Until their attention is drawn to it, many people do not realize how
>>> loud the background noise is in many common environments: buses, city
>>> streets, even the beach. I would be very skeptical of an e-book reader
>>> whose user interface depended on voice recognition, and considering
>>> the resources required, I would question the wisdom of offering it as
>>> an option.
>>
>> You could be right --- it's perhaps the least necessary of my 'specs',
>> and I would
>> likely jump at a reasonably priced device with all my other
>> 'requirements' but not that one
>> My tv remote shows me that you can get pretty far with a few buttons.
>> Yet ... you need to be able to do a search on author, title, keywords,
>> etc.
>> I wonder if a requirement that you talk "into the open e-book",
>> combined with a few strategically placed pickup microphones, couldn't
>> substantially filter out the background? Or, how about a wireless,
>> clip-on
>> microphone? Of course, at the beach, I'll have to clip it onto my hat :-)
>> Surely the acoustic engineers can solve this little problem.
>>
>> regards,
>> alan
>>
>>
>>
>>
>
>

On Sat, 28 May 2005 16:27:25 -0700, "Alan"
<info@optioncity.REMOVETHIS.net> wrote:

>However, the need to have multiple pages that you can physically
>'turn' seems to me entirely psychological unless I am missing something.
> I don't feel that need at all.

I can assure you that some other people do (I am one), and I suspect
that even more will resist giving up that capability even if they do
not need it.

When I'm reading a book, I often flip back through the last several
pages to compare what I'm reading to something I recently read, review
an important concept being built upon, etc. Sometimes I flip forward
for similar purposes. Sometimes I insert Post-it sheets as bookmarks
in a few places or a few dozen places. All of those things are more
awkward with an e-book, which makes them distracting and makes reading
more difficult. Sometimes I actually need to look at two pages at
once, for example to see a graph or table while reading text that
relates to it, and with an e-book that is simply impossible.

One's experience depends very much on what one is reading, of course.
For fiction or popular nonfiction, these limitations would matter much
less, if they mattered at all. Fiction or popular nonfiction
constitutes the majority of most people's reading. But most people
seem to have no particular interest in using e-books under any
conditions, if they have even noticed that such things exist. Their
motivation to adopt the new technology is limited. They have little
interest in annotating their reading, or in full-text searchability.
The economic advantage of an e-book over a mass-market paperback is
much less compelling than its advantage over an expensive scholarly
work or technical book. Under some conditions the economic advantage
goes the other way. If you go to the beach, for example, which would
you rather risk losing or damaging: a $5 paperback, or a $100 reader
loaded with a $1 electronic novel?

>Yet ... you need to be able to do a search on author, title, keywords, etc.
>I wonder if a requirement that you talk "into the open e-book",
>combined with a few strategically placed pickup microphones, couldn't
>substantially filter out the background? Or, how about a wireless, clip-on
>microphone?...
>Surely the acoustic engineers can solve this little problem.

It's a well-known problem. Plenty of acoustic engineers have worked on
it, and continue to do so. Noise canceling microphones can do
remarkable things, but speech recognition is a very demanding
application, and current technology has significant limitations.
You'll see what I mean if you cruise the Web and look for users'
comments on doing speech recognition with the microphones that are
built into current portable computers. I don't doubt that the
technology will improve, but the improvements are likely to be
incremental and slow.

Natural-language speech recognition is only really practical with a
noise canceling microphone positioned just to the side of the
speaker's mouth (usually on a boom attached to a headset). "Array
microphones" can function at distances of up to a couple of feet, but
they don't work as well, they probably wouldn't fit on a small
portable device, and they typically cost hundreds of dollars.

For limited-vocabulary speech recognition the requirements are more
relaxed, but I don't have a sense of how much so. And in any case, if
you want to use speech recognition to recognize authors' names, search
text, etc., you are no longer talking about a limited vocabulary!

I imagine this type of function being performed by something much
closer to a tablet PC, with handwritten input, communicating with the
bound e-book I described before through some type of wireless
connection.

My email address is LLM041103 at earthlink dot net.

I want to touch on a point that I missed in my previous message.

Natural-language speech recognition programs rely heavily on
word-sequence frequencies to achieve accuracy. For example, if the
program hears you say something that sounds like "I doughnut," it is
likely to conclude that you said "I do not," because that is a much
more likely sequence of words.

When you contemplate using speech recognition to enter authors' names,
search strings, etc., you have not only passed beyond the scope of
limited-vocabulary speech recognition; you have passed beyond the
scope of natural-language speech recognition, too. You are asking the
software to recognize speech that is not only drawn from an unlimited
vocabulary, but also lacks reliable word-frequency clues. This is
beyond the capability of any currently available commercial software,
even running on a full-fledged personal computer with a high-quality
boom microphone, used by an experienced operator under nearly ideal
conditions.

My email address is LLM041103 at earthlink dot net.

I like how you touched that one up.
Those are very good points.
-James

"Jonathan Sachs" <xxxxxxx@earthlink.not> wrote in message
news:qf7i91tdt1t47kh8j3g3bjvbfag45422h2@4ax.com...
>I want to touch on a point that I missed in my previous message.
>
> Natural-language speech recognition programs rely heavily on
> word-sequence frequencies to achieve accuracy. For example, if the
> program hears you say something that sounds like "I doughnut," it is
> likely to conclude that you said "I do not," because that is a much
> more likely sequence of words.
>
> When you contemplate using speech recognition to enter authors' names,
> search strings, etc., you have not only passed beyond the scope of
> limited-vocabulary speech recognition; you have passed beyond the
> scope of natural-language speech recognition, too. You are asking the
> software to recognize speech that is not only drawn from an unlimited
> vocabulary, but also lacks reliable word-frequency clues. This is
> beyond the capability of any currently available commercial software,
> even running on a full-fledged personal computer with a high-quality
> boom microphone, used by an experienced operator under nearly ideal
> conditions.
>
> My email address is LLM041103 at earthlink dot net.

"Jonathan Sachs" <xxxxxxx@earthlink.not> wrote in message
news:7r0i91hkius7q5s39bp021kp040a3nn9v0@4ax.com...
> On Sat, 28 May 2005 16:27:25 -0700, "Alan"
> <info@optioncity.REMOVETHIS.net> wrote:
>
>>However, the need to have multiple pages that you can physically
>>'turn' seems to me entirely psychological unless I am missing something.
>> I don't feel that need at all.
>
> I can assure you that some other people do (I am one), and I suspect
> that even more will resist giving up that capability even if they do
> not need it.
>
> When I'm reading a book, I often flip back through the last several
> pages to compare what I'm reading to something I recently read, review
> an important concept being built upon, etc. Sometimes I flip forward
> for similar purposes. Sometimes I insert Post-it sheets as bookmarks
> in a few places or a few dozen places. All of those things are more
> awkward with an e-book, which makes them distracting and makes reading
> more difficult. Sometimes I actually need to look at two pages at
> once, for example to see a graph or table while reading text that
> relates to it, and with an e-book that is simply impossible.

I do these things, too, and I agree that a properly designed e-book should
make them *easier*, not harder. That's why it's compelling to me.
First, assume for sake of argument that my ideal book exists with
perhaps a a good voice recog. and certainly *two* facing 6 x 9 pages,
which now seems a critical requirment to me.

Then, showing *any two* pages side by side would be easier than
a physical book. You simply announce "show pages 124 and 186".
Or, press a button to 'hold' the current page on the left, while you press
another button to flip forward or backward through pages that display
on the right.

Post-it notes: you announce "mark this page" or "mark the sentence starting
with ... " or press a "mark this page" button or use a tablet pen to
underline
something you want to remember. Later, when you come back to the
book, you can ask to see the marked pages or passages, or see new pages
that have been created that highlights of the text surrounding your
underlined areas.
I think software opens up so many more possibilities here.

For example, I have been doing some research lately involving the
probabilistic
concept of 'local time'. Not only would I like to electronically put post-it
notes
in a given book that discusses this, I would like to ask the book to
electronically bring up *all* the other books in my library that discuss the
same subject. This is quite tedious with my real books (they are scattered
all over my house), but would be vastly more convenient with the envisioned
device. Then, I could put a given page on hold, and flip through other
pages in other books on the facing display page.

Another problem, for me anyway, is that even when I have found what I need
in my own library, I
don't necessarily remember where I found it the next time. My e-book
would remember all these successful searches for me, say stored under the
'local time' search label. Indexing and local searching is becoming a big
deal for
me, and I have become a big fan of desktop search. I would want my e-book
to know about/implement/talk with/ those applications besides
having access to global searching on the net.

regards,
alan

"Jonathan Sachs" <xxxxxxx@earthlink.not> wrote in message
news:qf7i91tdt1t47kh8j3g3bjvbfag45422h2@4ax.com...
>I want to touch on a point that I missed in my previous message.
>
> Natural-language speech recognition programs rely heavily on
> word-sequence frequencies to achieve accuracy. For example, if the
> program hears you say something that sounds like "I doughnut," it is
> likely to conclude that you said "I do not," because that is a much
> more likely sequence of words.
>
> When you contemplate using speech recognition to enter authors' names,
> search strings, etc., you have not only passed beyond the scope of
> limited-vocabulary speech recognition; you have passed beyond the
> scope of natural-language speech recognition, too. You are asking the
> software to recognize speech that is not only drawn from an unlimited
> vocabulary, but also lacks reliable word-frequency clues. This is
> beyond the capability of any currently available commercial software,
> even running on a full-fledged personal computer with a high-quality
> boom microphone, used by an experienced operator under nearly ideal
> conditions.

I disagree that these terms necessarily come from open-ended unlimited
vocabularies.
First, book author names, titles, isbns, etc are from large, well-organized
*finite* collections of terms,
essentially "books in print". Next, academic authors of preprints or
journal articles are, in principle, collectible
into similar finite collections.

When you get to general "search phrases", you're right, but I would hope to
train
my ebook listener on the technical terms that I search most often for, as
again, this is
a limited list, often repeated.

If, in the end, certain searches required a tablet pen input to recognize
the
term, so be it.

regards,
alan

On Sat, 28 May 2005 19:45:57 -0700, "Alan"
<info@optioncity.REMOVETHIS.net> wrote:

>First, book author names, titles, isbns, etc are from large, well-organized
>*finite* collections of terms,
>essentially "books in print". Next, academic authors of preprints or
>journal articles are, in principle, collectible
>into similar finite collections.

I think you are misinterpreting the term "limited vocabulary," which
has a fairly specific meaning into speech-recognition field. Any
speech recognition system recognizes a vocabulary that is "limited" in
the sense of being finite. A limited-vocabulary system recognizes only
a few thousand words -- perhaps 30,000 at most, and the larger the
vocabulary, the poorer the performance. The practical limit for
speaker-independent recognition, without substantial user training,
and under somewhat adverse conditions, is much lower than that.

Books in Print (the American edition) is currently an eight-volume
set. I don't know how many distinct words and names are in the
listings, but I suspect that it far exceeds 20,000.

I think you are also underestimating the importance of word-sequence
clues, which I mentioned in an earlier message. Without these clues,
speech recognition is frankly just awful, even under conditions
approaching ideal. If you say an isolated word, like "a," there is no
telling whether the software will hear you correctly, or hear "any,"
"and," "AA," "they," or something even more far-fetched. Recognition
tends to be much better for longer words, like "parliamentary" or
"idiopathic," but that does not help with connecting words, or with
proper names, which tend to be highly irregular and homonymic. (Am I
Jonathan Sachs, Sachse, Sacks, or Saks, or if your pronunciation is
not perfect, Sack, Stack, or Satz? Without contextual information, the
software hasn't a clue.)

My email address is LLM041103 at earthlink dot net.

On Sat, 28 May 2005 19:31:48 -0700, "Alan" wrote:

>> When I'm reading a book, I often flip back through the last several
>> pages... flip forward... insert Post-it sheets as bookmarks...
>
>I do these things, too, and I agree that a properly designed e-book should
>make them *easier*, not harder. That's why it's compelling to me....
>
>Post-it notes: you announce "mark this page" or "mark the sentence
>starting with ... " or press a "mark this page" button or use a tablet pen to
>underline something you want to remember. Later, when you come back
>to the book, you can ask to see the marked pages or passages...

You've described several intelligently conceived features for making
these operations easier for an e-book user to perform. The problem is
that while you can make these operations much easier than they would
be on a poorly designed e-book, you can't make them nearly as easy as
they are with a physical book.

The reason for this is that the e-book is an artificial construct,
with artificial features and rules. Everything you do with it,
starting with how you load it and turn it on, must be _learned_, and
then must be practiced as a learned skill.

In contrast, a physical book shares most of the properties of any
other physical object. It has size, texture, and weight. It has parts
which are connected to each other in certain ways which determine the
ways they interact. Our brains are partly hard-wired to deal with
these properties, and are partly conditioned to deal with them from
the time we start interacting with the physical world, a few weeks
after birth. Thus the skills that we need to perform basic operations
on a book -- to pick it up and hold it, to turn the pages, to place
and find bookmarks -- are either instinctive, or are so deeply
conditioned that they might as well be instinctive.

Even if you learn a skill so well that it becomes automatic, it still
takes more "bandwidth" than one that is instinctive or conditioned.
Thus, while you may feel that you can "turn the page" on an e-book
just as easily as on a paper book, it's not exactly true. It may be
true under good conditions, but under stress -- when you're tired, or
when you're trying to keep three other things in short-term memory at
the same time -- the greater difficulty is reflected in your
performance.

The real problem is for the casual user, who does not use the gadget a
lot, and is not strongly motivated to learn it. That person may never
learn to use it well enough to become comfortable with it, and then
will say, "Yeah, I tried an e-book, but it was no good; you can't set
bookmarks." And for him, it's true, even though it's not true for the
gadget itself.

This is not "the fault of the user, not the product"; it's just the
way people think. No, it's not logical. Technology lovers are
not-logical too, in the same ways, just about different things.

You're correct that e-books can support sophisticated features like
look-up that are difficult with conventional books, and which
sophisticated readers will intuitively value. My point is not that
those features are impractical or useless, but that devices which
offer them will still have trouble gaining acceptance unless they do
the simple stuff about as easily as conventional books, and about as
well. And that, ironically, the simplest stuff is the most difficult
for an e-book to do in a satisfactory way.

This is why I favor the concept of a bound e-book, which can be
handled (although not loaded) without need for any learned skills at
all. Here's a link that shows what I mean. Note the paragraph that
begins, "For applications requiring more rapid and direct electronic
update..."

http://www2.parc.com/hsl/projects/gyricon/


My email address is LLM041103 at earthlink dot net.

> I think you are misinterpreting the term "limited vocabulary," which
> has a fairly specific meaning into speech-recognition field. Any
> speech recognition system recognizes a vocabulary that is "limited" in
> the sense of being finite. A limited-vocabulary system recognizes only
> a few thousand words -- perhaps 30,000 at most, and the larger the
> vocabulary, the poorer the performance. The practical limit for
> speaker-independent recognition, without substantial user training,
> and under somewhat adverse conditions, is much lower than that.

Jonathan,

It sounds like my ideal device needs to be highly constrained as far
as the speech recoginition goes. Thinking out loud, suppose I told it
I mostly read science and science-fiction and this reduced the
books-in-print
universe to say 20,000 items (an item being a title and author, with of
course more
unique titles than authors). Could a speech recognition system learn a
universe of
words of that size, so that it could recognize my pronunciation,
by hearing another computer program pronounce them all?

I am just trying to imagine the best way to train this system, if that is
possible
at all, with minimal input from the user. How would you do it?

regards,
alan

On Sun, 29 May 2005 10:36:39 -0700, "Alan" wrote:

> It sounds like my ideal device needs to be highly constrained as far
>as the speech recoginition goes. Thinking out loud, suppose I told it
>I mostly read science and science-fiction and this reduced the
>books-in-print
>universe to say 20,000 items (an item being a title and author, with of
>course more
>unique titles than authors). Could a speech recognition system learn a
>universe of
>words of that size, so that it could recognize my pronunciation,
>by hearing another computer program pronounce them all?
>
>I am just trying to imagine the best way to train this system, if that is
>possible
>at all, with minimal input from the user. How would you do it?

First, a caveat: I am not a technical expert in this area. I have a
lot of technical background, and I have relied on speech recognition
software for most of my computer use for about three years, because
carpal tunnel syndrome limits my ability to use a keyboard. In the
process I have picked up a good deal of technical lore, but when you
ask very specific questions, I'm not qualified to answer.

That said, a vocabulary of 20,000 items is certainly within the range
traditionally considered to be limited vocabulary. That's a big help.
It's also a big help if the system may require training to recognize
each user's speech. (I have been assuming otherwise, but I can't offer
cogent reasons for that.) Whether the system you envision is
technically feasible or not, or will become so in the foreseeable
future... I'm not even qualified to guess.

As a user of Dragon NaturallySpeaking, I'm accustomed to thinking of a
"vocabulary item" as a single word, or at most a two- or three-word
phrase. To give an example in your proposed subject area, the
vocabulary would contain "Kim," "Stanley," and "Robinson," and
hopefully the word-sequence frequency data would show that "Kim
Stanley Robinson" is a common sequence of words. The user might add
"Kim Stanley Robinson" as a distinct vocabulary item, improving
recognition for that name, but would never add "Years of Rice and
Salt, by Kim Stanley Robinson" as a single item. I don't know whether
current speech recognition systems would respond well to such a long
entry, or could be adapted to do so, or _should_ be adapted to do so.

One more caution: while it is easy for users to add words or phrases
or lists of them to a vocabulary, it takes a great deal of labor to
build a new vocabulary and the word-sequence frequency tables and
other arcana that accompany it. The person who does the work must be
trained in computational linguistics _and_ must be thoroughly familiar
with the internals of the speech recognition product in question. For
any given product, there may be a dozen qualified vocabulary builders
in the world! That is part of the reason why the medical and legal
versions of Dragon NaturallySpeaking each sell for a premium of about
$500 over the standard "Professional" version.

This poses a problem for the type of specialized vocabulary you are
contemplating: it may simply be unaffordable to create specialized
vocabularies for any but a small number of highly lucrative markets.

My email address is LLM041103 at earthlink dot net.

Thanks for the insights into speech recognition, Jonathan.
Very enlighterning for me.

regards,
alan