Dragon Naturally Speaking

Discussion in 'Computer Information' started by John, Nov 11, 2004.

  1. John

    John Guest

    Hello.

    I have Dragon Naturally Speaking but have a little problem.

    I have set myself up as a user without any trouble. However I also
    wish to have a transcription from a recorded sound file on my computer
    of an American Talk Show host.

    Is there anyway I can train the system to recognise his accent etc of
    this host and transcribe accurately?

    When you first set up a new user there is an option to record the
    words from a wav file as your dictation source, however, you have to
    repeat a certain story back to the system which is already in there.
    E.g. 3001 The Final Odyssey, Alices Adventures in Wonder Land, Charlie
    and the Chocolate Factory etc...

    This would be a problem because I would have to find all the
    individual words the Talk Show host says to put them all together for
    the story for that to work, and there are quite a few paragraphs of
    text for those stories.

    Is there any way I can import my own text into the system and train
    from that? This way, I would be able to type up a couple of
    paragraphs that the Talk Show host actually says in the wav file, and
    then I would simply be able to press play and it would all work.

    If it is possible to do that please let me know. Otherwise I may have
    to just base it on my own voice, and then there will be lots of
    inaccuracies as I am British.

    Thanks for any help on this.

    John
    John, Nov 11, 2004
    #1
    1. Advertising

  2. John

    Joe Guest

    I can't help you and I am not even sure anyone can but I have a question for
    you. How well does it do at recognizing your voice?
    I bought Via Voice about three years back and it was horrible. I spent
    the better portion of a day reading to my computer to train the program and
    it still stunk. I tried to dictate one email with it and gave up as even
    though I was a slow typist I could still just type faster than I could
    correct all the mistakes the program made. It got the wrong word altogether
    in many cases and rarely picked the right one out of the same word (to, too,
    or two / their, there, or they're).
    Did you make your post here with it if so how many mistake were there or
    how many mistakes would there have been if you had? If these products have
    gotten better I would not mind trying one again but the one I tried was a
    waste of money.
    Joe
    "John" <> wrote in message
    news:...
    > Hello.
    >
    > I have Dragon Naturally Speaking but have a little problem.
    >
    > I have set myself up as a user without any trouble. However I also
    > wish to have a transcription from a recorded sound file on my computer
    > of an American Talk Show host.
    >
    > Is there anyway I can train the system to recognise his accent etc of
    > this host and transcribe accurately?
    >
    > When you first set up a new user there is an option to record the
    > words from a wav file as your dictation source, however, you have to
    > repeat a certain story back to the system which is already in there.
    > E.g. 3001 The Final Odyssey, Alices Adventures in Wonder Land, Charlie
    > and the Chocolate Factory etc...
    >
    > This would be a problem because I would have to find all the
    > individual words the Talk Show host says to put them all together for
    > the story for that to work, and there are quite a few paragraphs of
    > text for those stories.
    >
    > Is there any way I can import my own text into the system and train
    > from that? This way, I would be able to type up a couple of
    > paragraphs that the Talk Show host actually says in the wav file, and
    > then I would simply be able to press play and it would all work.
    >
    > If it is possible to do that please let me know. Otherwise I may have
    > to just base it on my own voice, and then there will be lots of
    > inaccuracies as I am British.
    >
    > Thanks for any help on this.
    >
    > John
    >
    >
    Joe, Nov 11, 2004
    #2
    1. Advertising

  3. John

    Robert Baer Guest

    Joe wrote:
    >
    > I can't help you and I am not even sure anyone can but I have a question for
    > you. How well does it do at recognizing your voice?
    > I bought Via Voice about three years back and it was horrible. I spent
    > the better portion of a day reading to my computer to train the program and
    > it still stunk. I tried to dictate one email with it and gave up as even
    > though I was a slow typist I could still just type faster than I could
    > correct all the mistakes the program made. It got the wrong word altogether
    > in many cases and rarely picked the right one out of the same word (to, too,
    > or two / their, there, or they're).
    > Did you make your post here with it if so how many mistake were there or
    > how many mistakes would there have been if you had? If these products have
    > gotten better I would not mind trying one again but the one I tried was a
    > waste of money.
    > Joe
    > "John" <> wrote in message
    > news:...
    > > Hello.
    > >
    > > I have Dragon Naturally Speaking but have a little problem.
    > >
    > > I have set myself up as a user without any trouble. However I also
    > > wish to have a transcription from a recorded sound file on my computer
    > > of an American Talk Show host.
    > >
    > > Is there anyway I can train the system to recognise his accent etc of
    > > this host and transcribe accurately?
    > >
    > > When you first set up a new user there is an option to record the
    > > words from a wav file as your dictation source, however, you have to
    > > repeat a certain story back to the system which is already in there.
    > > E.g. 3001 The Final Odyssey, Alices Adventures in Wonder Land, Charlie
    > > and the Chocolate Factory etc...
    > >
    > > This would be a problem because I would have to find all the
    > > individual words the Talk Show host says to put them all together for
    > > the story for that to work, and there are quite a few paragraphs of
    > > text for those stories.
    > >
    > > Is there any way I can import my own text into the system and train
    > > from that? This way, I would be able to type up a couple of
    > > paragraphs that the Talk Show host actually says in the wav file, and
    > > then I would simply be able to press play and it would all work.
    > >
    > > If it is possible to do that please let me know. Otherwise I may have
    > > to just base it on my own voice, and then there will be lots of
    > > inaccuracies as I am British.
    > >
    > > Thanks for any help on this.
    > >
    > > John
    > >
    > >


    As i understand it, there are 2 types of speech to text programs.
    One requires training, and "recognizes" only that person.
    The other might be considered "generic" in that no training is
    required, and "recognizes" "anyone" within certain limitations.
    I have heard that neither type (for the PC) has ever been reliable or
    routinely useable.
    The qualification is on the second type; it works as long as the
    requirements are met.
    For example, you may have called a few companies, including the loco
    (pun intended) version of Ma Bell and were prompted to speak answers.
    In those cases, the responses (in each case) were rather limited, and
    counting of pauses (in the word) and consonants could be used (ie:
    ignore the words, per se).
    For the PC version, the "vocabulary" starts undefined and one must
    speak in a rather formal and precise manner (that is what i heard,
    anyway).
    Robert Baer, Nov 12, 2004
    #3
  4. John

    Wizard Guest

    Dragon is as good as it gets and works very well (after training.)

    Robert Baer wrote:
    >
    > Joe wrote:
    > >
    > > I can't help you and I am not even sure anyone can but I have a question for
    > > you. How well does it do at recognizing your voice?
    > > I bought Via Voice about three years back and it was horrible. I spent
    > > the better portion of a day reading to my computer to train the program and
    > > it still stunk. I tried to dictate one email with it and gave up as even
    > > though I was a slow typist I could still just type faster than I could
    > > correct all the mistakes the program made. It got the wrong word altogether
    > > in many cases and rarely picked the right one out of the same word (to, too,
    > > or two / their, there, or they're).
    > > Did you make your post here with it if so how many mistake were there or
    > > how many mistakes would there have been if you had? If these products have
    > > gotten better I would not mind trying one again but the one I tried was a
    > > waste of money.
    > > Joe
    > > "John" <> wrote in message
    > > news:...
    > > > Hello.
    > > >
    > > > I have Dragon Naturally Speaking but have a little problem.
    > > >
    > > > I have set myself up as a user without any trouble. However I also
    > > > wish to have a transcription from a recorded sound file on my computer
    > > > of an American Talk Show host.
    > > >
    > > > Is there anyway I can train the system to recognise his accent etc of
    > > > this host and transcribe accurately?
    > > >
    > > > When you first set up a new user there is an option to record the
    > > > words from a wav file as your dictation source, however, you have to
    > > > repeat a certain story back to the system which is already in there.
    > > > E.g. 3001 The Final Odyssey, Alices Adventures in Wonder Land, Charlie
    > > > and the Chocolate Factory etc...
    > > >
    > > > This would be a problem because I would have to find all the
    > > > individual words the Talk Show host says to put them all together for
    > > > the story for that to work, and there are quite a few paragraphs of
    > > > text for those stories.
    > > >
    > > > Is there any way I can import my own text into the system and train
    > > > from that? This way, I would be able to type up a couple of
    > > > paragraphs that the Talk Show host actually says in the wav file, and
    > > > then I would simply be able to press play and it would all work.
    > > >
    > > > If it is possible to do that please let me know. Otherwise I may have
    > > > to just base it on my own voice, and then there will be lots of
    > > > inaccuracies as I am British.
    > > >
    > > > Thanks for any help on this.
    > > >
    > > > John
    > > >
    > > >

    >
    > As i understand it, there are 2 types of speech to text programs.
    > One requires training, and "recognizes" only that person.
    > The other might be considered "generic" in that no training is
    > required, and "recognizes" "anyone" within certain limitations.
    > I have heard that neither type (for the PC) has ever been reliable or
    > routinely useable.
    > The qualification is on the second type; it works as long as the
    > requirements are met.
    > For example, you may have called a few companies, including the loco
    > (pun intended) version of Ma Bell and were prompted to speak answers.
    > In those cases, the responses (in each case) were rather limited, and
    > counting of pauses (in the word) and consonants could be used (ie:
    > ignore the words, per se).
    > For the PC version, the "vocabulary" starts undefined and one must
    > speak in a rather formal and precise manner (that is what i heard,
    > anyway).
    Wizard, Nov 12, 2004
    #4
  5. On Thu, 11 Nov 2004 20:06:47 +0000, John <> wrote:

    >Hello.
    >
    >I have Dragon Naturally Speaking but have a little problem.
    >
    >I have set myself up as a user without any trouble. However I also
    >wish to have a transcription from a recorded sound file on my computer
    >of an American Talk Show host.
    >
    >Is there anyway I can train the system to recognise his accent etc of
    >this host and transcribe accurately?
    >
    >When you first set up a new user there is an option to record the
    >words from a wav file as your dictation source, however, you have to
    >repeat a certain story back to the system which is already in there.
    >E.g. 3001 The Final Odyssey, Alices Adventures in Wonder Land, Charlie
    >and the Chocolate Factory etc...
    >
    >This would be a problem because I would have to find all the
    >individual words the Talk Show host says to put them all together for
    >the story for that to work, and there are quite a few paragraphs of
    >text for those stories.
    >
    >Is there any way I can import my own text into the system and train
    >from that? This way, I would be able to type up a couple of
    >paragraphs that the Talk Show host actually says in the wav file, and
    >then I would simply be able to press play and it would all work.
    >
    >If it is possible to do that please let me know. Otherwise I may have
    >to just base it on my own voice, and then there will be lots of
    >inaccuracies as I am British.
    >
    >Thanks for any help on this.
    >
    >John
    >


    I have upgraded every version since the first in search of the Holy
    Grail and despite those who proclaim that it has "arrived", it has
    in fact not arrived and I suspect the newest version is no different.
    I have trained and trained and trained, bought high priced mics, etc
    etc and it still types "for" when I mean "fore" (as just one of a
    multitude of examples. This is difficult technology.

    It appears that Wizard disagrees with this opinion but believe me, as
    a lawyer a good VR system would be a Godsend and it just ain't there
    yet. I figure 5 more years...
    John H. Power, Nov 13, 2004
    #5
  6. John

    gg Guest

    MA Bell's voive recogition system costs at least $60,000.
    Voice recognition has arrived... for $60,000 to $120,000.
    How much did you all say you spent?
    gg, Nov 14, 2004
    #6
  7. John

    gg Guest

    But all seriousness aside....
    Have you tried patching an inexpensive graphic-equilizer between the
    source (voice, wav file player) and the input port of your computer?
    The idea is to create a "tone" Dragon can better decipher. This is
    not the end-all solution. By tone I mean consonants that are
    distinctive and that really "pop". As a starting setting on your
    graphic-equilizer raise the equalizer's treble and mid range knobs as
    high as they can go and set bass knob at zero or below zero. Adjust
    them until Dragon gives the best results.

    For the radio guy if you don't want to buy an equilizer just adjust
    the treble, bass and mid-range knobs on your radio before you record.
    If you already have a wav file then adjust tone settings (treble, mid
    and bass) on the equalizer then make a dupe.

    Summarizing, the voice setup goes in the direction of the arrow.

    ======>
    Voice to mike to graphic-equalizer to computer-input-jack to Dragon.

    The radio setup goes in the direction of the arrow.
    Radio to equalizer to wave-file-recorder to input-jack to Dragon.
    gg, Nov 14, 2004
    #7
  8. John

    Guest

    On Sun, 14 Nov 2004 18:24:52 GMT, (gg) wrote:

    >MA Bell's voive recogition system costs at least $60,000.
    >Voice recognition has arrived... for $60,000 to $120,000.
    >How much did you all say you spent?


    Now I believe that to be true. This is why I think we are still at
    least 5 years out. If the technology really worked now everyone would
    have it and use it. The reason they do not is because it does not...
    , Nov 15, 2004
    #8
  9. John

    Deborah Guest

    Hi everyone,

    I noticed that some people asking questions about how well the
    voice-recognition works and thought I'd share my two cents. I happen
    to really enjoy using the voice-recognition software, what let's
    rephrase that. In comparison to having to use notetakers and writers
    this is a much better choice.

    I use Dragon NaturallySpeaking preferred. I've been using this
    program for over 10 years and have watched enormous transformation
    over the past years. a big factor for successful recognition is a
    computer with a high-speed processor, significant amount of RAM, a
    good microphone, and learning how to speak in a coherent matter. the
    more words you can string together at a time to more successful your
    recognition will be. For example, my last three sentences had no
    errors but my first sentence had several. Learning how to speak for
    dictation was not easy as I'm a Long Island girl. Anyone who knows
    people from Long Islandbe rather fast. This does not work well with
    dictation as when speaking fastwords or syllables tend to be garbled.
    I notice now almost all my errors argues her errors not the program.
    I must say this is only true for the 7.3 version which is at least
    three or four times faster than the seven .oh version (I left this
    mistake in place to show you a error which was technically a dragon
    error but was contributed by myself since I paused in the midst of the
    sentence.)

    for me the hardest technique to try to acquire, actually I am still
    trying, is the ability to say a full sentence at one time. This is
    not something that is easy to do in my areana. I'm finishing my law
    degree and many of the sentences we write have a great deal of meat
    in each portion. Being able to say the sentence in his attire day
    before dictation is almost impossible. I focused more on trying to
    say 10 or so words at once. When following this philosophy as well as
    speaking and enunciating each word the recognition and speed is quite
    impressive. If reading out of a textbook I can easily get 70 words or
    more per minute.

    Also, something I've learned to increase my speed is to train each
    word in the correction menu. In Dragon when using the "correct that"
    command a menu box appears that allows you to correct the word. One
    option also allows you to train the new word. I have found doing so
    significantly decreases in this recognition of this word in the
    future. I've only had on several occasions where the word and error
    reappeared more than once. It was not the fault of the software but
    that of my underlying accent. :) (I even used a command for this.) I
    don't have a very strong accent but some words are affected by it and
    I must train dragon with my different pronunciation.

    I'm using a Pentium 4 with a CPU of a 2.4mhz and 512 ram. With this
    system the software program runs rather quickly and can even play
    MP3s, iTunes, and stream music while dictating. Do note in order to
    do this you must use a USB microphone, I use a digital signal
    processing microphone from planatronics.

    I'm really excited about the prospects of the 8.0 version. For those
    readers interested, this is the preferred version, the middle level
    program.

    I'll be really happy when the voice dictation can work with Firefox
    like it works with Internet Explorer. I truly disliked the latter
    program and will be a very happy camper when Dragon can asked us all
    of the links via dictation as a canned in IE. For those of you
    unfamiliar with this, win in IE you can use Dragon to select links by
    simply saying "click and the first word of the link". If more than one
    exists, number tags will appear by each and the user chooses which
    link they desire.

    Please note in order to use this program successfully you truly do
    need a good processor with a strong CPU speed otherwise a processing
    speed will become quite frustrating. If you don't have the latter
    actually suggest waiting until you have the necessary hardware. As a
    person who was forced unto the use of Dragon because of disabilities I
    didn't have such a choice of. I can't describe the levels of
    frustration I encountered. To be honest its quite amazing that a
    computer never actually went through my bedroom window. :)

    As an fyi, I owe my undgrad, graduate and soon-to-be law degree to the
    use of Dragon. I actually conduct all law school exams with Dragon and
    use it for writing all papers and homework assignments. Without Dragon
    would not have been able to succeed in Law school since my disability
    significantly limits any writing or typing abilities.

    I noticed someone mentioned Via Voice. I tried it awhile back and
    truly hated the program. However, this was sometime ago so he no
    longer familiar with its engineering. The methods for corrections
    were more tedious and time-consuming than with Dragon. To be honest
    almost all my experiences with Dragon. Especially during the latter
    years as each upgrade has brought significantly increased accuracy and
    speed. For example must last paragraph I only had one error and it
    was mostly my fault as the word was slurred.

    Hope this helps!

    Sorry to write such a long letter, I promise to take a step off of the
    soapbox. Take care!
    Deborah, Nov 22, 2004
    #9
  10. On 21 Nov 2004 17:30:17 -0800, (Deborah) wrote:

    >Hi everyone,
    >
    >I noticed that some people asking questions about how well the
    >voice-recognition works and thought I'd share my two cents. I happen
    >to really enjoy using the voice-recognition software, what let's
    >rephrase that. In comparison to having to use notetakers and writers
    >this is a much better choice.
    >
    >I use Dragon NaturallySpeaking preferred. I've been using this
    >program for over 10 years and have watched enormous transformation
    >over the past years. a big factor for successful recognition is a
    >computer with a high-speed processor, significant amount of RAM, a
    >good microphone, and learning how to speak in a coherent matter. the
    >more words you can string together at a time to more successful your
    >recognition will be. For example, my last three sentences had no
    >errors but my first sentence had several. Learning how to speak for
    >dictation was not easy as I'm a Long Island girl. Anyone who knows
    >people from Long Islandbe rather fast. This does not work well with
    >dictation as when speaking fastwords or syllables tend to be garbled.
    >I notice now almost all my errors argues her errors not the program.
    >I must say this is only true for the 7.3 version which is at least
    >three or four times faster than the seven .oh version (I left this
    >mistake in place to show you a error which was technically a dragon
    >error but was contributed by myself since I paused in the midst of the
    >sentence.)
    >
    >for me the hardest technique to try to acquire, actually I am still
    >trying, is the ability to say a full sentence at one time. This is
    >not something that is easy to do in my areana. I'm finishing my law
    >degree and many of the sentences we write have a great deal of meat
    >in each portion. Being able to say the sentence in his attire day
    >before dictation is almost impossible. I focused more on trying to
    >say 10 or so words at once. When following this philosophy as well as
    >speaking and enunciating each word the recognition and speed is quite
    >impressive. If reading out of a textbook I can easily get 70 words or
    >more per minute.
    >
    >Also, something I've learned to increase my speed is to train each
    >word in the correction menu. In Dragon when using the "correct that"
    >command a menu box appears that allows you to correct the word. One
    >option also allows you to train the new word. I have found doing so
    >significantly decreases in this recognition of this word in the
    >future. I've only had on several occasions where the word and error
    >reappeared more than once. It was not the fault of the software but
    >that of my underlying accent. :) (I even used a command for this.) I
    >don't have a very strong accent but some words are affected by it and
    >I must train dragon with my different pronunciation.
    >
    >I'm using a Pentium 4 with a CPU of a 2.4mhz and 512 ram. With this
    >system the software program runs rather quickly and can even play
    >MP3s, iTunes, and stream music while dictating. Do note in order to
    >do this you must use a USB microphone, I use a digital signal
    >processing microphone from planatronics.
    >
    >I'm really excited about the prospects of the 8.0 version. For those
    >readers interested, this is the preferred version, the middle level
    >program.
    >
    >I'll be really happy when the voice dictation can work with Firefox
    >like it works with Internet Explorer. I truly disliked the latter
    >program and will be a very happy camper when Dragon can asked us all
    >of the links via dictation as a canned in IE. For those of you
    >unfamiliar with this, win in IE you can use Dragon to select links by
    >simply saying "click and the first word of the link". If more than one
    >exists, number tags will appear by each and the user chooses which
    >link they desire.
    >
    >Please note in order to use this program successfully you truly do
    >need a good processor with a strong CPU speed otherwise a processing
    >speed will become quite frustrating. If you don't have the latter
    >actually suggest waiting until you have the necessary hardware. As a
    >person who was forced unto the use of Dragon because of disabilities I
    >didn't have such a choice of. I can't describe the levels of
    >frustration I encountered. To be honest its quite amazing that a
    >computer never actually went through my bedroom window. :)
    >
    >As an fyi, I owe my undgrad, graduate and soon-to-be law degree to the
    >use of Dragon. I actually conduct all law school exams with Dragon and
    >use it for writing all papers and homework assignments. Without Dragon
    >would not have been able to succeed in Law school since my disability
    >significantly limits any writing or typing abilities.
    >
    >I noticed someone mentioned Via Voice. I tried it awhile back and
    >truly hated the program. However, this was sometime ago so he no
    >longer familiar with its engineering. The methods for corrections
    >were more tedious and time-consuming than with Dragon. To be honest
    >almost all my experiences with Dragon. Especially during the latter
    >years as each upgrade has brought significantly increased accuracy and
    >speed. For example must last paragraph I only had one error and it
    >was mostly my fault as the word was slurred.
    >
    >Hope this helps!
    >
    >Sorry to write such a long letter, I promise to take a step off of the
    >soapbox. Take care!


    I appreciate your take on this but I too have been using the program
    since its inception, also with high powered equipment, so we are
    together on that. Where we differ is that I have been practicing law
    for 25 years and it would be a godsend if this program delivered as
    represented but it just does not. Perhaps you have gotten it to
    perform miracles but I have been unable to. I will tell you that once
    you begin practicing law, your reputation will rise or fall on the
    quality of your output. One or 2 misinterpreted words that are not
    caught can create a bad impression so just make sure you proofread
    every word after you have dictated. I have always found that it took
    more time to correct the stupid mistakes e.g. "to" instead of "two"
    than it was worth. I still think we've got 5 more years to go before
    it's performance becomes acceptable...
    John H. Power, Nov 25, 2004
    #10
  11. John

    Deborah Guest

    John,

    I definitely agree with you, Dragon does have plenty of room for
    improvement. I do understand your remarks about misinterpreted words.
    I actually lost a significant number of points on a torts exam when a
    'command' appeared in mid sentence dividing my list of elements. The
    prof. actually stopped reading the paragraph. Actually this had caused
    some hurdles during exams because I did not receive the proper
    extension of time on exams. Part of the comprimise with the Dean led
    to a policy whereby corrections were made after the exam. Imagine
    trying to read a con law exam and making all the necessary corrections
    in under 30 minutes. And now imagine doing so without a mouse. :)
    This problem also reared it's ugly face while clerking for an EEOC
    judge.

    Up until recently, I quite frequently had misinterpreted words. This
    significantly decreased with version 7.3 and my current plantronics
    microphone. The latter made a signifcant difference. Also, most other
    mics inplanted words when I breathed deeply/heavily. This no longer
    happens.

    I want to take a moment to clarify a previous statement in which I
    said this program is a "godsend". It's all relative. Would I rather
    type. Heck yeah. I'm an amazing typist. In fact, I was abled to type
    lecture notes verbatium. When timed, I was consistenly clocked in the
    mid 130's

    My disability prevents me from typing or writing (all the joints in my
    fingers, hands and wrist dislocate/sublux upon movement.) I honestly
    don't believe I would be here, at the same position in my educational
    endeavors without it. I had an extremely difficult time trying to
    work with writers in graduate school. My lowest grades correlated
    with writers. I actually preferred Dragon Dictate with, on a good day,
    10 words per minute to the writer. So for me, despite some drawbacks,
    this program has saved the day. And, now I'm impressed with it's
    performance.

    FYI, last night my voice became very hoarse from dictation and a cold.
    Recognition was still in the high 90'.

    Take care,
    Deb
    >
    > I appreciate your take on this but I too have been using the program
    > since its inception, also with high powered equipment, so we are
    > together on that. Where we differ is that I have been practicing law
    > for 25 years and it would be a godsend if this program delivered as
    > represented but it just does not. Perhaps you have gotten it to
    > perform miracles but I have been unable to. I will tell you that once
    > you begin practicing law, your reputation will rise or fall on the
    > quality of your output. One or 2 misinterpreted words that are not
    > caught can create a bad impression so just make sure you proofread
    > every word after you have dictated. I have always found that it took
    > more time to correct the stupid mistakes e.g. "to" instead of "two"
    > than it was worth. I still think we've got 5 more years to go before
    > it's performance becomes acceptable...
    Deborah, Dec 3, 2004
    #11
  12. John

    Deborah Guest

    J I have upgraded every version since the first in search of the Holy
    > Grail and despite those who proclaim that it has "arrived", it has
    > in fact not arrived and I suspect the newest version is no different.
    > I have trained and trained and trained, bought high priced mics, etc
    > etc and it still types "for" when I mean "fore" (as just one of a
    > multitude of examples. This is difficult technology.


    There are a couple of techniques which I've used to clear up these
    types of errors.

    (1) I ran a program within the accuracy center program which scans
    previously written documents. The engine scans these documents to
    ascertain the occurance and pattern of words so that it may predict
    the appropriate "for".

    (2) I've noticed the program tends to decrease usage of the
    inappropriate version of the word after one or two corrections. Again
    learning to predict more accurately. Personally, I go about this by
    saying "select fore" and chose the correct "four". In most
    situations, I've only needed to make this correction one time.

    (3) One of the most accurate means for decreasing this problem is to
    increase the number of words said at a time. I encounter this problem
    3 or 4 more times when dictating only a small group of words; 3 -5
    words. This problem is almost non-existent when dictating long
    sentences at a time; 10 or more words. If you watch the anchor box
    you'll notice Dragon adjusts its intrepration of the dictation as the
    input of words increases.


    Of course this is easier said than done. Especially when writing legal
    documents. I've found a happy medium between all of the above,
    especially the latter two. I'm still learning how to form long
    sentences prior to dictation of sentence. It's no small task. Lately,
    I've focused more on dictating half a sentence at a time. This has
    worked rather well.

    Take care,
    Deb

    BTW You wouldn't believe the accuracy I obtained when dictating case
    briefs. It was rather easy to dictate long sentences with the rule of
    law before my eyes. Now only if exams were that easy. Or for that
    matter, the case I'm working on right now.
    Deborah, Dec 3, 2004
    #12
  13. John

    Robert Baer Guest

    Deborah wrote:
    >
    > J I have upgraded every version since the first in search of the Holy
    > > Grail and despite those who proclaim that it has "arrived", it has
    > > in fact not arrived and I suspect the newest version is no different.
    > > I have trained and trained and trained, bought high priced mics, etc
    > > etc and it still types "for" when I mean "fore" (as just one of a
    > > multitude of examples. This is difficult technology.

    >
    > There are a couple of techniques which I've used to clear up these
    > types of errors.
    >
    > (1) I ran a program within the accuracy center program which scans
    > previously written documents. The engine scans these documents to
    > ascertain the occurance and pattern of words so that it may predict
    > the appropriate "for".
    >
    > (2) I've noticed the program tends to decrease usage of the
    > inappropriate version of the word after one or two corrections. Again
    > learning to predict more accurately. Personally, I go about this by
    > saying "select fore" and chose the correct "four". In most
    > situations, I've only needed to make this correction one time.
    >
    > (3) One of the most accurate means for decreasing this problem is to
    > increase the number of words said at a time. I encounter this problem
    > 3 or 4 more times when dictating only a small group of words; 3 -5
    > words. This problem is almost non-existent when dictating long
    > sentences at a time; 10 or more words. If you watch the anchor box
    > you'll notice Dragon adjusts its intrepration of the dictation as the
    > input of words increases.
    >
    > Of course this is easier said than done. Especially when writing legal
    > documents. I've found a happy medium between all of the above,
    > especially the latter two. I'm still learning how to form long
    > sentences prior to dictation of sentence. It's no small task. Lately,
    > I've focused more on dictating half a sentence at a time. This has
    > worked rather well.
    >
    > Take care,
    > Deb
    >
    > BTW You wouldn't believe the accuracy I obtained when dictating case
    > briefs. It was rather easy to dictate long sentences with the rule of
    > law before my eyes. Now only if exams were that easy. Or for that
    > matter, the case I'm working on right now.


    ....and i have noticed that the accuracy of each of your responses here
    has increased by leaps and bounds.
    I presume you have experimented with mis-proe-nownced words to improve
    conversion accuracy, and a specific voice inflection at the end of a
    string of words in the "middle" of a sentence - so as to help prevent
    the software from making that string a sentence.

    Sounds like the training is making it better than the "spell checker"
    that M$ Nerd has...
    Robert Baer, Dec 3, 2004
    #13
    1. Advertising

Want to reply to this thread or ask your own question?

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.
Similar Threads
  1. Kevin Wallis

    Dragon Naturally Speaking

    Kevin Wallis, Jul 15, 2003, in forum: Computer Support
    Replies:
    0
    Views:
    590
    Kevin Wallis
    Jul 15, 2003
  2. Jack the Ripper

    Dragon Naturally Speaking 7.0

    Jack the Ripper, Sep 29, 2003, in forum: Computer Support
    Replies:
    1
    Views:
    673
    Robert Larson
    Sep 29, 2003
  3. Joshuall

    Dragon Naturally Speaking

    Joshuall, Aug 31, 2005, in forum: Computer Support
    Replies:
    2
    Views:
    510
    old jon
    Sep 1, 2005
  4. vanbar6

    Dragon naturally speaking

    vanbar6, Oct 5, 2006, in forum: Computer Information
    Replies:
    1
    Views:
    392
    Tee Jay
    Oct 5, 2006
  5. anna
    Replies:
    4
    Views:
    2,654
    thanatoid
    Apr 24, 2006
Loading...

Share This Page