Results 1 to 14 of 14
  1. #1
    2 Star Lounger
    Join Date
    Dec 2009
    Location
    West Midlands, UK
    Posts
    132
    Thanks
    116
    Thanked 1 Time in 1 Post

    Voice recognition

    I find Google Chrome’s voice recognition for entering website names fantastic. I would like to be able to similarly dictate notes on my tablet and desktop – is the software available for download as a standalone program and/or tablet app. My needs do not justify buying a program such as Dragon Speaking Naturally but I cannot find a “free” or open source alternative – any help please.

  2. The Following User Says Thank You to mike21 For This Useful Post:

    Chris Cooper (2014-07-24)

  3. #2
    3 Star Lounger
    Join Date
    Dec 2009
    Location
    UK
    Posts
    348
    Thanks
    17
    Thanked 73 Times in 61 Posts
    Windows has had speech recognition built-in since Vista (although admittedly it wasn't great at that time). However, WSR (Windows Speech Recognition) within Windows 7 is vastly improved over Vista and I've found it to be extremely useful. (I have no experience of WSR in Windows 8.x.)

    WSR is easy to setup and use. For example, using Windows 7, just click on the Start orb and type speech. The top search result should be Windows Speech Recognition.
    wsr.jpg
    Click to enlarge

    A wizard will help you set up your headset/microphone and take you through a tutorial to get you started. After many years of using both Dragon NaturallySpeaking (DNS) since v7, including the Professional versions, and WSR since Vista the best advice I can give you is to persevere and practise. Be aware too that to get the most out of WSR (and DNS) you will have to learn and remember a large number of commands. Tip - In WSR, say What can I say for a list of commands appropriate to the context.

    Google Chrome's voice recognition is limited functionality and tuned for a specific purpose. DNS and WSR have wide-ranging uses and abilities, both to convert speech to text and to control the computer. As a result both benefit enormously from training the program to recognise your voice. What this means is... don't expect miracles immediately as results improve enormously with regular use.

    I also use Windows Speech Macros to extend WSR's functionality. These macros (similar but less capable, unfortunately, than DNS's scripting) allow you to create composite actions, i.e. a number of actions all triggered by a spoken word or phrase. As an example, here's a Google that macro:

    Code:
    <speechMacros>
    
      <!-- Written by Glen Shires   08-08-08 -->
     
      <!-- Select text in any window and say "Google that" or "Ebay that" or "Yahoo Answers that" -->
      <!-- and the search results will appear in your default web browser. -->
    
      <!-- A complete list of the web-sites that you can say are at the bottom of this file, -->
      <!-- and you can easily add more! -->
    
      <!-- Alternatively, you can search for any text you dictate:  -->
      <!--    For example, say "Google for 'summer olympics'".  -->
      <!-- However, this free-form dictation is sometimes error-prone, and won't work for words not in your speech dictionary. -->
    
      <!-- So a third alternative is to search on text appearing anywhere in the current document window (without needing to manually select it).  -->
      <!--    For example, say "Google my 'waitForDisambiguation'"  -->
      <!-- Which probably isn't in your speech dictionary, but works when this document is open.  -->
      <!-- Note that this alternative only works for documents in editable windows (like Word or Notepad). -->
      <!-- If there are multiple occurences, it will ask you to select one (for example, say "One OK"). -->
    
      <command priority="3">
        <listenFor>[website] ?on that</listenFor>
        <listenFor>[website] ?for that</listenFor>
        <listenFor>[website] ?on this</listenFor>
        <listenFor>[website] ?for this</listenFor>
        <script language="VBScript">
          <![CDATA[ 
          Application.SendKeys("{250 WAIT}{{CTRL}}c{250 WAIT}") 
          that = Application.clipboardData.GetData("text") 
          Application.Run("{[website.url]}"+that)
        ]]>
        </script>
      </command>
    
      <command priority="2">
        <listenFor>[website] for [anytext...]</listenFor>
        <listenFor>[website] on [anytext...]</listenFor>
        <run command="{[website.url]}{[anytext...]}"/>
      </command>
    
      <command priority="1">
        <listenFor>[website] my [textInDocument]</listenFor>
        <emulateRecognition waitForDisambiguation="15">select {[textInDocument]}</emulateRecognition>
        <emulateRecognition>{[website]} that</emulateRecognition>
      </command>
    
      <command priority="0">
        <listenFor>[website] *</listenFor>
        <setTextFeedback style="warning">What was that?</setTextFeedback>
      </command>
    
      <!-- List of the websites you can say... -->
    
      <listenForList name="website" propname="url">
        <item propval="http://www.google.com/search?q=">google</item>
        <item propval="http://images.google.com/images?q=">google images</item>
        <item propval="http://www.google.com/maps?q=">google maps</item>
        <item propval="http://www.google.com/news?q=">google news</item>
        <item propval="http://www.google.com/products?q=">google shopping</item>
        <item propval="http://search.yahoo.com/search?p=">yahoo</item>
        <item propval="http://images.search.yahoo.com/search/images?p=">yahoo images</item>
        <item propval="http://news.search.yahoo.com/news/search?p=">yahoo news</item>
        <item propval="http://shopping.yahoo.com/search?p=">yahoo shopping</item>
        <item propval="http://local.yahoo.com/results?stx=">yahoo local</item>
        <item propval="http://answers.yahoo.com/search/search_result?p=">yahoo answers</item>
        <item propval="http://search.live.com/results.aspx?q=">live search</item>
        <item propval="http://www.google.com/search?q=define%3A">dictionary</item>
        <item propval="http://thesaurus.reference.com/browse/">thesaurus</item>
        <item propval="http://en.wikipedia.org/wiki/Special:Search?search=">wikipedia</item>
        <item propval="http://search.ebay.com/search/search.dll?satitle=">ebay</item>
        <item propval="http://sfbay.craigslist.org/search/sss?query=">craigslist</item>
      </listenForList>
    
    </speechMacros>
    More info:
    • The built-in Windows Help system has lots of advice about WSR whilst Google is your friend for both WSR and Windows Speech Macros.
    • There's a Yahoo group specifically for Windows Speech Macros that may be worth joining.
    • Vocola is an open source macro scripting add-on to WSR.
    • Neither DNS Home nor DNS Premium offer the same scripting ability as DNS Professional. DNS Pro is much more expensive.

    Hope this helps...
    Last edited by Rick Corbett; 2014-07-24 at 17:52.

  4. The Following 4 Users Say Thank You to Rick Corbett For This Useful Post:

    bcoop (2014-07-24),Chris Cooper (2014-07-24),Mike Feury (2014-08-16),wavy (2014-07-16)

  5. #3
    2 Star Lounger
    Join Date
    Dec 2009
    Location
    West Midlands, UK
    Posts
    132
    Thanks
    116
    Thanked 1 Time in 1 Post
    Thanks - have printed out your advice and will have a go.

  6. #4
    New Lounger
    Join Date
    Apr 2013
    Posts
    3
    Thanks
    0
    Thanked 1 Time in 1 Post
    While the retail price for Dragon Naturally Speaking is rather high, it can often be found at substantial discount prices (at least here in the US). There are often posts on slickdeals.net with prices ranging from $20-50 US. I picked up version 12, with a headset, for $20 last year. If you take the time to go through the process of training DNS to recognize your voice and manner of speaking, then it does remarkably well at writing what you say. It also integrates into Word and Outlook, so it will write directly into those applications. Finished composing an email in Outlook? Just say "Send Message" and DNS sends it on its way. Yes, I'm a fan.

  7. The Following User Says Thank You to HarryH3 For This Useful Post:

    Chris Cooper (2014-07-24)

  8. #5
    Lounger
    Join Date
    May 2010
    Location
    Cornwall UK
    Posts
    45
    Thanks
    3
    Thanked 11 Times in 10 Posts
    Thanks to mike21 for the question and Rick Corbett and HarryH3 for their encouraging responses.

    Only yesterday I was asked by a very partially sighted client if I thought voice recognition software would work for him. As his computer skills are quite limited, and my past experience of earlier versions of DNS being a bit hit and miss, I said I thought he may be better to stay with the screen magnifier for as long as he could.

    Following all your good advice I'll see if I can find him a reasonably priced later version of DNS and give it a whirl on his system.

    Best regards, Chris
    Industrial electrical engineer, running a system building/repair business in Cornwall, UK, for the last 14 years.

    Built my first computer in 1978 - in the days when you had to hand-solder in all the components!
    Until Windows 8, I thought we were still moving forwards!

  9. #6
    New Lounger
    Join Date
    Apr 2013
    Posts
    3
    Thanks
    0
    Thanked 1 Time in 1 Post
    Chris, I too had tried a MUCH earlier version of DNS and found it extremely lacking. But that was around 1996 or so and I was using a laptop with either a 486 or early Pentium CPU. That first impression kept me from spending any more money on DNS for many years. I'm glad that I finally gave it another chance. The most difficult part of using it is learning to "say" all of the punctuation so DNS will insert the commas, periods, colons, etc. into the document.

    I think your client will find it very useful. Be aware that a new version was just announced so there should be some excellent prices on version 12 now that version 13 has become available.

  10. #7
    4 Star Lounger petesmst's Avatar
    Join Date
    Dec 2009
    Location
    Cape Town, South Africa
    Posts
    508
    Thanks
    15
    Thanked 23 Times in 15 Posts
    I have used Dragon Naturally Speaking for several years (and heave recently updated to Version 13). I am using it to type this, in fact. Once you have mastered it, you have very little need for the keyboard or mouse!
    (My Setup: 3,70GHz Intel Core i7-4820K CPU; MSI Military Class iii X79A-GD45 Plus Motherboard; Win 8.1 Pro (64 bit); 16GB RAM; SAMSUNG SD840 PRO SSD (6GB/SATA III); Seagate 2TB Barracuda SATA6G HDD; GeForceGTX 760 2GB Graphics Card; Office 2013 Prof (32-bit); MS Project 2013 (32-bit); Acronis TI 2014 Premium, NIS 2014, etc). (UEFI-booted). WD My Book 3 1TB USB External Backup Drive)

  11. #8
    Star Lounger
    Join Date
    Jul 2011
    Posts
    83
    Thanks
    31
    Thanked 3 Times in 3 Posts
    Quote Originally Posted by petesmst View Post
    I have used Dragon Naturally Speaking for several years (and heave recently updated to Version 13). I am using it to type this, in fact. Once you have mastered it, you have very little need for the keyboard or mouse!
    Pity it can't always spell "have"...

  12. #9
    3 Star Lounger
    Join Date
    Dec 2009
    Location
    UK
    Posts
    348
    Thanks
    17
    Thanked 73 Times in 61 Posts
    Chris - One thing to point out with DNS is that you need an Internet connection to activate each installation.

    On the plus side, you can install DNS on two devices at a time (as it's licensed 'per user'). The documentation assumes a desktop and laptop but doesn't appear to check. As a result I've had recent DNS versions running on 2 laptops at the same time with no problems. If you do end up installing it on 2 devices, I suggest you change the location of the user profile within DNS so they both point to the same location, e.g. network, shared drive or (fast) USB stick, so just one profile is used for storing voice files, voice training data, etc.

    Also, make a regular backup of the user's DNS profile. Unfortunately, corruption of the user's DNS profile is not unknown and it can take ages to build up a new profile from scratch. Having a backup avoids all the tedium of having to train DNS again.

    The amount of available RAM is very important. Nuance show a recommendation for 2Gb minimum for DNS v13. I would suggest doubling that. We used to double the amount of minimum recommended RAM - from 2Gb to 4GB - going back to v10 after finding that this was one of the best performance improvements for DNS. Also, try to avoid running it on devices with low cost processors, e.g. entry-level Celeron-based laptops. DNS requires more oomph than most other applications in order to convert speech to text quickly. The last thing you want is a considerable delay before speech appears converted on screen.

    Have a look at Nuance's own Tips and Tricks for using DNS.

    Final tip - Keep an eye out for patches and Service Packs. In my experience (including supporting 40+ DNS users in a business environment for 12 years) each new version often brings a small number of quirks and irritations that are sorted out with a subsequent Service Pack.

    Hope this helps...
    Last edited by Rick Corbett; 2014-07-24 at 15:55.

  13. #10
    4 Star Lounger petesmst's Avatar
    Join Date
    Dec 2009
    Location
    Cape Town, South Africa
    Posts
    508
    Thanks
    15
    Thanked 23 Times in 15 Posts
    @Trev: You have proved an extremely important point (at my expense!): Dragon 13 claims to be 15% more accurate than version 12 and that it is very accurate straight out of the box. I installed it less than a day before my above post. I had carried out no voice training within the newly-installed program (in contrast to the extensive training I had done with version 12). It is quite clear that even this new and improved Dragon 13 has problems with my South African accent! So, thanks for pointing out the error; I must now clearly go through the built-in training routines and ensure that I do not use a program I'm trying to praise in a forum such as this, and make a fool of myself! Once one builds confidence in dictation software it is extremely easy to fall into the trap of not proofreading what one has dictated before committing it to the Internet by clicking on "send" or "post"! (By the way, this post has also been dictated using Dragon 13 without any prior voice training -and note that this time I skilfully avoided use of the word "have"-!)
    (My Setup: 3,70GHz Intel Core i7-4820K CPU; MSI Military Class iii X79A-GD45 Plus Motherboard; Win 8.1 Pro (64 bit); 16GB RAM; SAMSUNG SD840 PRO SSD (6GB/SATA III); Seagate 2TB Barracuda SATA6G HDD; GeForceGTX 760 2GB Graphics Card; Office 2013 Prof (32-bit); MS Project 2013 (32-bit); Acronis TI 2014 Premium, NIS 2014, etc). (UEFI-booted). WD My Book 3 1TB USB External Backup Drive)

  14. #11
    3 Star Lounger
    Join Date
    Dec 2009
    Location
    UK
    Posts
    348
    Thanks
    17
    Thanked 73 Times in 61 Posts
    @petesmst: One of the selling points of DNS 13 is that, unlike previous versions, initial voice training for the new version is no longer recommended... so it's a shame you've had this experience.

    Did you let DNS 'upgrade' your previous DNS profile or did you start from scratch? The reason I ask is because I came across this on the KnowBrainer forum:

    "A tiny smidgen of advice: While Dragon has successfully upgraded your Home Edition Ver. 12 profile to Ver. 13, hence the (V 13) suffix. If you have any vocabulary that you would like to preserve, from Ver. 12, you should click on the Vocabulary menu and choose Export custom word and phrase list. The reason we are making this recommendation is because you really shouldn't use an upgraded user profile. While upgraded profiles tend to be accurate, they also tend to be somewhat lethargic because they are not pure Ver. 13 user profiles. We recommend creating a brand-new user profile from scratch and reversing the process by importing your previously exported custom vocabulary." (Post marked 07/24/2014 07:56 PM)

    Hope this helps...
    Last edited by Rick Corbett; 2014-07-25 at 13:51.

  15. #12
    Lounger
    Join Date
    May 2010
    Location
    Cornwall UK
    Posts
    45
    Thanks
    3
    Thanked 11 Times in 10 Posts
    Many thanks to all for your very comprehensive advice on DNS. It sounds as if v13 has a lot going for it, particularly for someone with limited computer skills.

    I’m probably making a stick for my own back but I am going to recommend it to this new, partially sighted client – wish me luck!

    Cheers, Chris

    P.S. petesmst: I too have an anglo-saxon based regional accent so DNS will probable 'heave' me too, rather than 'have' me!
    Industrial electrical engineer, running a system building/repair business in Cornwall, UK, for the last 14 years.

    Built my first computer in 1978 - in the days when you had to hand-solder in all the components!
    Until Windows 8, I thought we were still moving forwards!

  16. #13
    3 Star Lounger
    Join Date
    Dec 2009
    Location
    UK
    Posts
    348
    Thanks
    17
    Thanked 73 Times in 61 Posts
    @Chris Cooper - If you're going to make the leap then here's another couple of tips:

    • Use a headset that connects via USB (or wireless) rather than one with 3.5 mm audio jack connectors. (This is from many years of experience trying to get good results from built-in audio chipsets via an analogue connection.)
    • All headsets are not equal and if you buy the wrong one then recognition will suffer. Check Nuance's Hardware Compatibility List for products that have been tested favourably with DNS.
    • Don't buy DNS Home. (This is a recommendation from the speech recognition experts on the KnowBrainer forum. A specific comment from one of their gurus is "We basically consider the Home Edition to be a throwaway speech recognition appetizer." [Posted 07/23/2014 10:25 PM])
    • Most improvements to DNS 13 appear to be in the area of controlling the PC/laptop rather than improvements to speech-to-text (again, according to the KnowBrainer forum). If you're more interested in speech-to-text then consider DNS 12... it should be available much cheaper.
    • There are many ways to assist users with disabilities. Consider the use of (relatively) simple scripting languages to automate tasks, e.g. using keyboard shortcuts. For example, instead of using DNS to automate the insertion of a user's home address when the phrase 'my address' is spoken (and recognised correctly ), consider that this can also be achieved by a very simple one-line script using AutoHotkey (see here and/or here). As another example, the following 2-line AutoHotkey script will hook into Windows 7's built-in Magnifier utility and allow zooming in and out using the Windows key plus mousewheel together.
      Code:
      LWin & WheelUp::SendInput #{NumpadAdd}
      LWin & WheelDown::SendInput #{NumpadSub}

    Hope this helps...

  17. The Following User Says Thank You to Rick Corbett For This Useful Post:

    Trev (2014-07-25)

  18. #14
    4 Star Lounger petesmst's Avatar
    Join Date
    Dec 2009
    Location
    Cape Town, South Africa
    Posts
    508
    Thanks
    15
    Thanked 23 Times in 15 Posts
    @Rick Corbett: Thanks for that. Yes, I instinctively decided not to convert my "old" DNS profile. I have, in fact found DNS 13 (by the way, I have Version 13 Premium) to be remarkably accurate "out of the box" (except for the "heave" / "have" issue!!). I spent some time this morning training the program and have no doubt this will improve things (as was the case with DNS 12).

    This post appeared "as it is" at first attempt, using DNS 13 Premium - it even got heave and have as spoken!.
    (My Setup: 3,70GHz Intel Core i7-4820K CPU; MSI Military Class iii X79A-GD45 Plus Motherboard; Win 8.1 Pro (64 bit); 16GB RAM; SAMSUNG SD840 PRO SSD (6GB/SATA III); Seagate 2TB Barracuda SATA6G HDD; GeForceGTX 760 2GB Graphics Card; Office 2013 Prof (32-bit); MS Project 2013 (32-bit); Acronis TI 2014 Premium, NIS 2014, etc). (UEFI-booted). WD My Book 3 1TB USB External Backup Drive)

  19. The Following User Says Thank You to petesmst For This Useful Post:

    Trev (2014-07-26)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •