Softpanorama

May the source be with you, but remember the KISS principle ;-)
Contents Bulletin Scripting in shell and Perl Network troubleshooting History Humor

Dragon Naturally Speaking

First of all let me say that we do not have good alternatives to Dragon NaturallySpeaking Home 12.0. This is pretty much the only game in town. In version 12 the program is fairly capable and at $50 affordable.

First of all let me say that we do not have good alternatives, so to a certain extent reviews, which mark the program less then three stars demonstrate low level of reviewer adaptability. In version 12 the program is fairly capable and at $50 affordable.

Speech recognition is a very complex task for machines to do. As a result they are not very good at it althouth there is substantial progress for the last 10 years (first of all due to the tremendous growth of capabilities of hardware). But it is unrealistic to expect for them to achieve the level of human recognition. That will not happen for a long time, probably never. Some ambiguities of English make correct recognition highly depending of understanding of the context, and this "understanding" part will never be present in computers.

But brute force approach proved to be "good enough". That means that the key factor in quality of recognition is the speed of your hardware. Dragon installed on the computer with 3.4 GHz i7 CPU and 16 GB memory running at 1600 MHz and having SSI disk generally will beat Dragon installed on a laptop with i5 or i3 CPU, 4GB of memory and low speed 5600 RPM drive. By how much is a good question, but here there is no replacement for displacement.

And you should not overload the computer with other CPU intensive tasks while dictating. For example, if you need to watch a video, do it on the second computer. Number of running programs does not matter much as long as they require human input to operate so you can have MS Word, Excel and several other programs opened simultaneously without a problem.

But again there is no replacement for displacement and you want better recognition you need to pay for it by buying better hardware: top of the line CPU and high speed memory are a must in this game. Like in real-time computer games, you lose if you have inferior computer.

Another important thing to understand is that a certain (and higher then in typing) amount of errors is inevitable. One sure way to improve accuracy is to learn artificial "dictating" style in which you make additional "unnatural" pauses between words and/or speak in short three-four words phases, making pause after each phrase.

In other words, it is a human, who needs to adapt to Dragon capabilities not vice versa ;-). And that implies that each user should print and read Dragon manual several times, especially part about improving the vocabulary. As dull as it sounds it helps to improve recognition quality. You can also experiment with settings.  For example, in Tools/Options/Misc settings you  can try to find optimal for your hardware "Speed vs. Accuracy" setting. And the it's not always the "highest" setting. I have more luck with the "fastest" setting.

One additional variable here is the microphone. I think USB microphone works slightly better with the program. If you have difficulties with quality, it might makes sense to buy a more expensive standalone USB microphone.  Microphone positioning is an additional important variable that affects quality of recognition.

Some words will be systematically "misrecognized" by the program. In this case you can train the program to correctly recognize them. The number of words you specifically train your computer to recognize by pronouncing them is also very important, as typically you dictate texts limited just to three-four topics and thus have a specific vocabulary that partially can be entered in the PC in your own voice. This is a time consuming job, but it substantially improves result with the "difficult" for the program words.

The other key for successful usage of the program is to low your expectations. Please understand that perfect recognition of human voice is impossible outside of some simplistic limited dictionary scenarios. The fact that the current version can recognize text slightly faster then a typical typist type and make at this speed acceptable number of mistakes if you are careful and train yourself to use a program (which requires at least a month of day-to-day usage) is simply amazing.

Actually working with the program I got tremendous respect for the capabilities of human brain, which solve this incredibly complex task so efficiently and almost effortlessly.


Top Visited
Switchboard
Latest
Past week
Past month

Old News ;-)

NUANCE Dragon NaturallySpeaking 12 Home  

Low price is about $45...
 

ref=dp_top_cm_cr_acr_txt

A review for new users and how to get the most out of Dragon, September 26, 2012

By

Charles Bittner "Disabled comedian & gamer!" (New England USA) - See all my reviews
(TOP 500 REVIEWER)    (VINE VOICE)    (REAL NAME)    

This review is from: Dragon NaturallySpeaking Home 12.0, English (DVD-ROM)

Customer review from the Amazon Vine™ Program (What's this?)

***UPDATE*** I find that Dragon NaturallySpeaking 12 is spot on in terms of getting my words right. I've only used it for a few days but as far as voice recognition goes it's great. I followed all the advice below (because it's the best way to get good dictation) I feel that this product only gets better ever iteration. And I only use it for dictation and not mouse commands etc.

The following is a guide to helping new users to voice dictation to understand how to get the most out of Dragon NaturallySpeaking. Now these principles apply to all voice dictation software. I will update this review with more of the new features and how Dragon NaturallySpeaking 12 runs when I've had time to use the software more.

(This whole review is done with Dragon I will not edit it at all)

As a quadriplegic standup comedian (okay I'm the cant stand up comic) I spent a lot of time doing voice dictation. I have been using Dragon naturally speaking since 1995 and I must say it is truly an amazing product and Dragon Dictate for Mac/Dragon NaturallySpeaking are only getting better. I want to give a basic overview on voice dictation and being successful.

One other thing to note I do not use Dragon for mouse movements and commands as I had the ability to use a touchpad mouse so I will not delve into those features in my review. This will merely focus on how you can use voice dictation effectively to receive near 100% accuracy. And how Dragon does it almost perfectly every time.

One thing you need to think about when you get into voice dictation is speaking in word clumps. You have to think about what you want to have Dragon type for you then you have to say those sentences in chunks (I hope that makes sense) most people who complain about voice dictation don't take the time to realize this.

In the following example I did edit the text when errors were made for continuity of thought.

[Example]
if I wanted to say "Is that the wheelchair that goes upstairs?" "No it is not but it goes downstairs one time really fast!"
The way I attacked this sentence was "Is that the wheelchair -pause- that goes upstairs question mark -pause- no it is not -pause- but it goes downstairs -pause- one time really fast exclamation point

Here is another one "what is the worst thing that a handicapped person can hear? Is it a doctor telling you you'll never walk again? Or someone saying you can't go with us because you are disabled? No the worst thing we can hear is someone in a crowd yell "run for your lives" at that point I know I am boned.

Here is how I approached saying that joke out loud I will put in a - where I pause
what is the worst thing-that a handicapped person-can hear question mark - is it a doctor-telling you-you'll never walk again question mark - or someone saying-you can't go with us-because you are disabled-question mark - no-the worst thing that we can hear-is someone in a crowd-yell-open quote- run for your lives-end quote- at that point-I know I'm boned period
[End example]

I hope that helps you again understanding how to speak in clumps of words remember to keep your word count between pauses at about 7 to 10. If you do this you will be successful in using voice dictation. If you put a microphone near your mouth and learned to speak in word clumps I say you will have 99% accuracy every time.

Without speaking a word clumps and thinking about what you want to say is when voice dictation has problems, here is an example of a long sentence I will speak this without stopping and pausing at certain points.

What I said: I am not disabled I am just very lazy and really I'm only in this thing for the parking.
What Dictate heard: I am not disabled I am just very lazy and really I'm only in this thing for the parking.

There you go it got a perfect 19 word sentence correct but most of the time keeping your clumps lower helps the program not make mistakes that is just what I have found.

How to train your Dragon: DO ALL THE TRAINING Is all I have to say! Sorry for yelling out better now. But really one thing I found super helpful in doing voice dictation is doing all of the training that you can. While you might not feel like reading a bunch of chapters from random books it really helps and should be done. I make a rule to do all the training the program provides so that it gets used my voice.

Now the microphone that comes with Dragon is decent and will get you almost 99% accuracy. Right now I'm using a tabletop microphone I'm sitting about 6-8 inches away from the microphone and still getting perfect accuracy! Truly amazing how far it's come in 15 years.

What I appreciate also about Dragon is its low resource usage. I mean as I use Dragon through the years it would talk a lot of resources i.e. CPU power and RAM space but now on even a low-cost computer will be fully capable of running this powerful program. I would happily recommend at least 2 GB of RAM though.

Till I Collapse by Eminem lyrics, this is unedited straight dictation. First line is dictation - (Second is the true verse to compare.)
--------------------
As soon as the verse starts I eat it at MC's heart - (Soon as a verse starts I eat it at MC's heart) * ERROR
what's he thinking? How not to go against me? Smart. - (what is he thinking? How not to go against me? Smart.) * PERFECT
And it's absurd how people hang on every word - (And its absurd how people hang on every word.) * PERFECT
I'll probably never get the props I feel I ever deserve - (I'll probably never get the props I feel I ever deserve) * PERFECT
but I'll never be served my spot is forever reserved - (But I'll never be served my spot is forever reserved) * PERFECT
if I ever leave Earth that would be the death of me first. - (If I ever leave earth that would be the death of me first.) * PERFECT
Visit my heart of hearts I know nothing could ever be worse. - (Cause in my heart of hearts I know nothing could ever be worse.) * ERROR
That's why I'm clever when I put together every verse - (That's why I'm clever when I put together every verse) * PERFECT
my thoughts are sporadic, I act like I'm inadequate - (My thoughts are sporadic, I act like I'm an addict) * ERROR
I rap like I'm addicted to smack like I'm Kim Mathers. - (I rap like I'm addicted to smack like I'm Kim Mathers.) * ERROR
the fact is I would rather sit back and bomb some rappers. - (The fact is I would rather sit back and bomb some rappers.) * PERFECT
So this is like a full-blown attack I'm launching Them - (So this is like a full blown attack I'm launching at them) * ERROR
the track is on some battling raps who wants some static - (The track is on some battling raps who want some static) * PERFECT
because I don't really think that the fact that I'm slim matters - (Cause I don't really think that the fact that I'm Slim matters) * ERROR
a plaque of platinum status is wack if I'm not the baddest. - (A plaque of platinum status is whack if I'm not the baddest.) * ERROR
--------------------

I would say that there were a few errors in that resuscitation of that rap but not bad, and good enough that it would save me about 30 minutes to type that okay maybe 15 but I do type one key at a time.

ONE BONUS OF VOICE DICTATION TO NOTE:
Is that voice dictation sometimes makes mistakes but those mistakes can really be great. So sometimes voice dictation will misunderstand what you say what the end result is better than or uses different words then you are going to ever use. For example the other day I went to use "definitely demonstrates" and Dragon burped out "deftly demonstrates" and in the context it's error was a better fit is so bonus when that happens! Keep that in mind voice dictation for the win!

So all in all I recommend Dragon to anyone with a disability for sure, and if you are dyslexic Dragon can really help you also. I think you'll find that this program will free you from the keyboard when it comes to entering tax into e-mails or documents web forums and the like.

I am Chuck Bittner and I approve this message LOL!

Version 12 is a DOWNGRADE! Grab Version 11 Before it is Sold Out  By SanjeevP TOP 500 REVIEWER

August 31, 2012

 When it comes to speech recognition software, Dragon is pretty much the only game in town. Ever since Dragon Naturally Speaking version 9, the accuracy of voice recognition has been very good.

The upgrade from Dragon Naturally Speaking version 9 to 10 improved the accuracy of voice recognition perceptibly; and even version 11 improved the voice recognition noticeably. But Nuance disabled the use of Dragon Naturally Speaking version 11 in Electronic Medical Record programs. They want to sell, Dragon Naturally Medical version, several times more expensive to the healthcare market. And the irony of all this is that despite costing several times more, Dragon Medical 11 does not work any better for electronic medical records than Dragon 10. You just feel being ripped off!

Having used Dragon Naturally Speaking version 12 for 4 weeks now, I think version 12 is actually a DOWNGRADE and not an upgrade because of following reasons.

1. SLOWER: I have been spending a lot of time watching the spinning dragon in the correction window on Intel i3 processor 10 Gb RAM computer. Sometimes 6 seconds for a correction.

2. BUGGY: The Dictation Box keeps popping up in a variety of programs even when you are directly dictating into the program and do not want it. For example, you are dictating a document in Microsoft Word and suddenly the Dictation Box pops up and all dictation starts going to the box instead of Microsoft Word document.

3. Many of the old formatting commands do not work, and you have to learn a whole set of formatting commands all over again. Nuance is claiming lot of new formatting commands but so far I have seen very few new usable commands.

4. It does not work for Electronic Medical Record programs.

5. It does not work on Remote Desktop programs like RDC, Logmein, RDW, JumpDesktop etc.

6. Releasing a program in late 2012, tailgated by Windows 8 release, that does not work well with Windows 8 and calling it an upgrade?

But version 12 does have a couple of advantages.

1. The repertoire of recording devices has increased.

2. Tutorial is nice, as of now only text, and no videos.

According to Nuance, the accuracy has improved by 20% but the improvement is hard to perceive in day to day use because Version 12 is slower. Slight improvement in accuracy for a slower program is not a trade off to call an upgrade. Version 12 is supposed to have social media commands, but I have not used those.

Based on my personal experience with recording or voice inputting devices on Dragon Naturally Speaking version 6 to 12, I found the following.

A. USB Microphones: give the best voice recognition. Andrea NC-181 VM USB High Fidelity Monaural USB Computer Headset (P-C1-1022300-1) is good value for money. And if you are dictating long documents or do dictation stop-n-go piecemeal then connection between microphone and computer is solid.

B. Bluetooth Microphones: Work fairly well, but not as good as USB microphones. PLANTRONIC Calisto Headset with USB Dongle - 81493-02 is very good. The biggest handicap with wireless microphones is that connection between the microphone and the computer keeps breaking and they do not work well for piecemeal stop-n-go and dictation. Every-time you start dictating after a long pause, you have to reestablish the connection.

C. Recording Devices: Sony ICD-MX20DR9 32MB Flash-Based Digital Voice Recorder & Sony ICD-SX712 Digital Flash Voice Recorder give excellent accuracy.

D. Smartphones: I have not had good experience with iOS devices like iPhone or iPod touch and the main problem is that their connection with computer keeps breaking off, in middle of dictation, which is very disruptive when you are dictating a long document or pause for 15-30 minutes. Getting a phone call or text in middle of dictation can be troublesome. iOS devices are probably OK for short snippets for Facebook posting, as shown in Demo, but not long document dictation. it is supposed to work with Android devices too, but I have not used it personally.

If you are already using Dragon version 10 or 11, then stick with it because you will lose a lot of functionality in version 12 without gaining much in accuracy.

If this is the first time you are buying Dragon and you do not plan to use voice recognition for electronic medical records directly in the EMR program or on remote desktop (like LogMeIn, RDC etc.) then go for version 12. Otherwise, stick with version 10 or 11 or grab Dragon NaturallySpeaking Premium 11, [Old Version] at a much lesser price. Version 11 has excellent accuracy, is fairly fast, and is not buggy but does not work with Electronic Medical Records. Dragon NaturallySpeaking 10 Preferred - Medium Box works in most electronic medical record programs and on less powerful computers but is less accurate than version 11. Dragon NaturallySpeaking 9 Preferred [OLD VERSION] works in all electronic medical record programs that I know off. If not at Amazon, you can find older versions at off the beaten path retailers online, very cheaply.

Record to Text ? By Barry Ritholtz

May 13th, 2010

Quick question:

I want to be able to record an interview, either in person or over the phone, then somehow have that recording converted to text.

That is Speech-to-Text, (not the other way around).

Google voice is only so-so for this; Plan B is to take MP3 files and have someone in India transcribe them for $10 per hour.

I’d love to find a software solution to this.

Any ideas?

  1. Mark E Hoffer Says:
    May 13th, 2010 at 5:06 pm

    http://www.nuance.com/naturallyspeaking/

    Nuance is the Market Leader.. http://www.nuance.com/naturallyspeaking/landing/small-business.asp

    many other Warez available http://clusty.com/search?input-form=clusty-simple&v%3Asources=webplus&query=Speech+to+Text+Software

    if one was interested in ‘Road-Testing’/ doing a compare and contrast, could be a worthwhile project for an intern..

    ~~~

    BR: I have Dragon Naturally Speaking — upgraded both my PC at the office, and the Mac at home, and the damned software thing insists on being retrained. Its frickin exhausting.

  2.  alfred e Says:
    May 13th, 2010 at 5:12 pm

    Check with NSA. They’ve been doing speech and voice recognition for some time. But if you want to get even better check into the semantic/intent analysis they have been funding at major universities.

    Perhaps all those commercial companies got their start there.

    As a last resort backtrack the work Lucent (now Alcatel-Lucent) has been doing. My recent info suggests it has been spun off in a separate corp, just like the micro-cameras they invented.

  3.  DL Says:
    May 13th, 2010 at 5:12 pm

    Ten bucks an hour isn’t such a bad deal; voice-recognition software isn’t going to be flawless anyway.
     

  4. Mark E Hoffer Says:
    May 13th, 2010 at 5:15 pm

    Arequipa,

    do this http://www.smartcode.com/downloads/voice-to-text.html , at least (:
     

  5. msaroff Says:
    May 13th, 2010 at 5:17 pm

    I would agree with Mr. Hoffer. Naturally speaking is the market leader in text to speech.

    You will have to clean it up afterwards, but it does a decent job.

    Note also: It is designed for a person to train the software for their personal use, so your results would likely be worse for a random person and no training, particularly if an accent is involved.
     

  6.  patient renter Says:
    May 13th, 2010 at 5:38 pm

    Unfortunately innovation in the speech recognition area has been stagnant for many years.

  7. Tbrander Says:
    May 13th, 2010 at 5:39 pm

    Barry, Windows,, all versions since XP have some pretty good speech recognition built-in,, worth a spin,, from my experience the built-in Windows speech recognition works about as well as the Dragon paid for product..

    In fact it is probably worth the experiment to try out the Google speech to text with the mp3 file.. Google keeps getting better and the batch interface may be way better,,,Google seems to have better non-trained recognition.. have not tried though.

  8. dolbydog Says:
    May 13th, 2010 at 5:55 pm

    Barry, there is a recent article on exactly why you should just send them to India for transcription.

    http://robertfortner.posterous.com/the-unrecognized-death-of-speech-recognition

    Best of luck.

  9. changja Says:
    May 13th, 2010 at 6:01 pm

    Try this, it may be more humorous than accurate, google voice has a voicemail to text feature. Call someone with google voice, pipe the message through and see how well it translates! Let me know if you need an invite.

  10.  Mark E Hoffer Says:
    May 13th, 2010 at 6:08 pm

    dolbydog,

    that’s a good article, should give insight into why Nuance is so focused on “Doctors”, and “Lawyers”..at the min..

  11.  subscriptionblocker Says:
    May 13th, 2010 at 6:09 pm

    Never used this stuff – but here’s a review of that dragonvoice. Seem to remember IBM was using it.

    http://www.consumersearch.com/voice-recognition-software

  12.  dancin Says:
    May 13th, 2010 at 6:13 pm

    My job is developing speech recognition systems for large companies. Unfortunately the technology just isn’t there yet for very accurate random untrained voice recognition. It’s very good when you have knowledge of what is going to be said, but for random speech in an uncontrolled setting, it’s still quite a ways off.

    If accuracy is important at all, transcription is your only option. If you just need the general gist of the conversation, the google speech to text is probably sufficient if you do it quickly after the interview so you can remember what was said and correct the bad recognitions.

  13.  subscriptionblocker Says:
    May 13th, 2010 at 6:24 pm

    http://en.wikipedia.org/wiki/Speech_recognition

    Suspect you’re just slightly ahead of the “good stuff”…probably on it’s way to a Walmart near you using off the shelf DSPs. Surprisingly (?), few must have bought those PC packages…so your first practical speech to text converter may appear embedded within a kids toy?

    Military
    [edit] High-performance fighter aircraft

    Some important conclusions from the work were as follows:

    Speech recognition has definite potential for reducing pilot workload, but this potential was not realized consistently.
    Achievement of very high recognition accuracy (95% or more) was the most critical factor for making the speech recognition system useful — *with lower recognition rates, pilots would not use the system*.
    More natural vocabulary and grammar, and shorter training times would be useful, but only if very high recognition rates could be maintained.

  14.  Myr Says:
    May 13th, 2010 at 6:31 pm

    Get one of our long term unemployed to do it for $5.

  15.  subscriptionblocker Says:
    May 13th, 2010 at 6:32 pm

    http://www.nch.com.au/scribe/

    Aussies usually make good solid stuff. Know nothing about this one, but these guys built the answering machine software I bought 7(?) years ago.

    I’m a hardware guy – so entrusting *anything* to software was very difficult. Runs on an old thinkpad 600E and never breaks.

    If you can’t get Microsofts latest OS to run without annoyance, find an old copy of Win 2K and do this to it:

    http://www.litepc.com/xplite.html

    Uses Microsofts own hidden built ins to rip out the bloat. MS becomes very stable.

  16.  subscriptionblocker Says:
    May 13th, 2010 at 6:34 pm

    http://www.nch.com.au/software/dictation.html

    PC audio is their specialty.

  17.  vachon Says:
    May 13th, 2010 at 6:36 pm

    Send it to me. I’ll do it for a large coffee. 

  18.  Jonathan Says:
    May 13th, 2010 at 6:37 pm

    If you have the interview in a courtroom, they record it all for you on the taxpayer’s dime!

  19.  Jonathan Says:
    May 13th, 2010 at 6:38 pm

    …and transcribe it too!

  20.  Evoo Kermartin Says:
    May 13th, 2010 at 6:51 pm

    What the hell is wrong with you Barry? It’s called HIRE A HOT TRANSCRIPTIONIST.

    f*cking guy

  21.  Barry Ritholtz Says:
    May 13th, 2010 at 7:01 pm

    Who needs THAT arounnd the office !

    More trouble — no thanks.

  22.  greg Says:
    May 13th, 2010 at 7:18 pm

    Barry..I’ll do it for $9.50/hr. Cdn, dollars and a signed copy of your book—-oh, and a mention in your blog.

  23.  rgc Says:
    May 13th, 2010 at 7:42 pm

    You really can’t find a transcriptionist in the USA who would do this for $10/hour? I find that hard to believe in this economy. If true, then this country truly is doomed.

  24.  lalaland Says:
    May 13th, 2010 at 7:53 pm

    er, there’s an app for that?

    I think it’s called jott but it probably sucks for anything more complicated than run fido run but what do I know…

  25.  gloppie Says:
    May 13th, 2010 at 8:14 pm

    Dragon / IBM is what’s used for half closed-captioning TV services, the other half being live trained operator using courtroom-style Stenotype to captions equipments.
    I installed a Dragon system and the tweak was to trash the original crappy mike that comes with, and use a decent omni dynamic mike with some compression and bandwidth limiting from 400 Hz. to around 8 KHz. It gets it right around 95% of the time.

  26.  panchog Says:
    May 13th, 2010 at 8:21 pm

    “Dragon dictation” app for iPhone or iPod Touch/iPad?

    It’s free & it’s pretty accurate.

  27.  Rehabengineer Says:
    May 13th, 2010 at 8:29 pm

    The Nuance products (speech to text) are speaker dependent so limited use in this situation. More speaker independent tech is coming out regularly-mostly for telecom voice access apps.

    Try you question at SpeechTek magazine. They are on top of the software apps for speech.

    Speech.Technology@emediapro.com

  28.  blu Says:
    May 13th, 2010 at 8:38 pm

    I haven’t tried it myself, but I know people that have used Amazon Mechanical Turk for this. You take your mp3, split it into short segments, and offer a small amount for each translation. You do each segment several times to avoid errors. It works out to be cheaper than the $10 per hour.

  29.  donna Says:
    May 13th, 2010 at 9:10 pm

    It hasn’t gotten any better in the 20 years since I worked for a voice recognition company? Huh. Would’ve thought we could have solved this one by now….

    Oh well, just like all that artificial intelligence software I worked on, I suppose! ;^)

    At least the Internet worked out well.

  30.  apikoros Says:
    May 13th, 2010 at 9:19 pm

    Before you go for DragonDictate, you might want to see Brad DeLong’s current post:

    http://delong.typepad.com/sdj/2010/05/the-beatings-will-continue-until-morale-improves-dragondictate-for-iphone-department.html

    I have no experience, so I have no further comment 

  31.  Mike in Nola Says:
    May 13th, 2010 at 9:48 pm

    I gave up on dragon years ago and haven’t heard that it’s gotten substantially better. I vote for India, or maybe someone wearing a sandwich board who’d do it for little more.

  32.  The MacDaddy Says:
    May 13th, 2010 at 9:51 pm

    I use a service called copytalk, you dial, play your tape, speak etc and you get an email back.

    http://www.copytalk.com/mobilescribe.po?

  33.  thehofa Says:
    May 13th, 2010 at 10:32 pm

    This is a pretty great VM service…

    http://www.simulscribe.com/

  34.  constantnormal Says:
    May 13th, 2010 at 10:48 pm

    filter out everyone who does not have first-hand experience and does not currently use software to do this … it’s a pretty small set of responders.

    Myself, I take this to mean that the technology has not yet arrived in a professional sort of way. When you think about all the peripheral recognition skills that go into speech recognition, and the sophisticated semantic analysis that goes into resolving homynyms, it’s not surprising. This is a VERY difficult problem.

    If software was able to accomplish speech recognition (with an acceptably low error rate), we would have perfect grammar checkers — and yet we do not.

    If a certain amount of errors are acceptable, go ahead an try out the best software packages recommended — but if you don’t want to have to manually review the result, playing the audio while you read it back, stick with a quick-response web service with a human being at the other end of the connection.

    Some things are still best done by the puny humans.

  35.  dcsos Says:
    May 14th, 2010 at 12:32 am

    Mac Speech Dictate
    is 100% improvement over dragon

    Don’t know if it’ll do untrained voices
    but its exact, where Dragon Dictate fails

  36.  Chris Says:
    May 14th, 2010 at 1:57 am

    Whoa, 10$ per hour? Don’t go to India, in Germany you will find lots of qualified people who work for a lot less.

  37.  Mike Says:
    May 14th, 2010 at 7:12 am

    I looked into this a few years ago and wasn’t really satisfied with price, quality and turnaround time so I abandoned the project. It will get better with time, but technology isn’t there yet.

    If you need a relatively quick turnaround and are willing to pay up, check out:
    http://www.speak-write.com

    You can phone in your dictation, upload recordings or download iPhone or Android apps for it and they get your work back to you in under 3 hours. They’ll charge you 1-2 cents per word.

  38.  Mr.E. Says:
    May 14th, 2010 at 7:22 am

    BR, sadly, dancin has it right.

    About a year ago I wanted to do essentially what you describe for a very large project involving customer feedback for a huge multinational corporation. We investigated and tried about every readily available option and in then end it had to be done the brute force way – transcription and even that was not as simple as it sounds. Speech-to-text conversion for random speech isn’t yet capable of dealing with the ginormous multitude of dialect variations. Even our transcribers had difficulty with the wide range of dialects seen in the U.S. alone, and when we threw in non-American native speakers it got really interesting. Add to it the complications of a recorded session and the loss of fidelity plus added noise and it becomes about near impossible. We even tried STT conversion as a first step and then have a transcriptionist “clean it up” thinking it would save time. The transcriptionists went nuts and all told us it would take them less time to just do it from scratch, brute force. The best deal we came up with was using college and talented high-school co-ops for about the same as what you quote for India. Sourcing your transcription locally will give you much better control over the product (meaning less do-over and polishing when you get it – and it WILL need polishing even with good transcription), and I suspect you can find some very capable talent via local education co-op programs.

  39.  tmmike Says:
    May 14th, 2010 at 8:17 am

    Try Amazon’s Mechanical Turk Community – for a reasonable rate you can have the audio transcribed several times and use the compare feature in a word processor to confirm an accurate transcription. You can set qualifications for those who want to do your work, and over time, you will develop a group of “turkers” who will do your work with the precision you want. Often these are college students who can’t work a fixed schedule or live in college towns with little available employment.
    https://www.mturk.com

    User Community
    http://turkers.proboards.com/index.cgi?

    Worth a few minutes of your time

  40.  carrottop Says:
    May 14th, 2010 at 9:23 am

    the industry leader (when i looked into it a long time ago) is NUANCE
    http://www.nuance.com/naturallyspeaking/products/product-comparison.asp

    i used to have their stock (NUAN), hard to sleep w/ a PE of 450….

  41.  pseudboy Says:
    May 14th, 2010 at 10:39 am

    India would be more like 2 bucks per hour. For $10, I’m sure you could find a ton of people in the US.

    BTW, the nuance/dragon software sucks.

    Seeking alpha puts up earnings call transcripts in a matter of a few hours. They are generally very accurate. I don’t know what technology they use but I’m sure you could pull a few strings to find out.

  42.  Deborah Says:
    May 14th, 2010 at 8:03 pm

    Dragon Speech Recognition Family of Products
    Turn Talk into Type

    Most people speak over 120 words per minute but type less than 40 words per minute. What if you could create email, documents and spreadsheets simply by speaking? What if you could control your PC just by talking to it? This includes launching applications, opening files, managing e-mail and working on the Web — all by voice.

    With speech recognition software from Nuance Communications, you can turn your voice into text three times faster than most people type. Just start talking, and the software will recognize your voice instantly, delivering up to 99% accuracy as soon as you get started. Accuracy will continually improve the more you use the software.

    It’s easy to get started with speech recognition, whether you’re using a PC or a Mac. Each edition of our speech recognition software delivers the same fast and accurate transcription of spoken words. But some editions include more advanced features to make interacting with your computer – regardless of whether it’s a PC or a Mac — easier than ever.

  43.  harryappenzeller Says:
    May 14th, 2010 at 11:45 pm

    amy goodman at democracy now ! has been providing very good quality rush transcripts of her interviews for years – you should ask her who she gets to do it… here’s a link to her site
    http://www.democracynow.org/
    and a link to her contact info which has a NYC phone
    http://www.democracynow.org/contact



Etc

FAIR USE NOTICE This site contains copyrighted material the use of which has not always been specifically authorized by the copyright owner. We are making such material available in our efforts to advance understanding of environmental, political, human rights, economic, democracy, scientific, and social justice issues, etc. We believe this constitutes a 'fair use' of any such copyrighted material as provided for in section 107 of the US Copyright Law. In accordance with Title 17 U.S.C. Section 107, the material on this site is distributed without profit exclusivly for research and educational purposes.   If you wish to use copyrighted material from this site for purposes of your own that go beyond 'fair use', you must obtain permission from the copyright owner. 

ABUSE: IPs or network segments from which we detect a stream of probes might be blocked for no less then 90 days. Multiple types of probes increase this period.  

Society

Groupthink : Two Party System as Polyarchy : Corruption of Regulators : Bureaucracies : Understanding Micromanagers and Control Freaks : Toxic Managers :   Harvard Mafia : Diplomatic Communication : Surviving a Bad Performance Review : Insufficient Retirement Funds as Immanent Problem of Neoliberal Regime : PseudoScience : Who Rules America : Neoliberalism  : The Iron Law of Oligarchy : Libertarian Philosophy

Quotes

War and Peace : Skeptical Finance : John Kenneth Galbraith :Talleyrand : Oscar Wilde : Otto Von Bismarck : Keynes : George Carlin : Skeptics : Propaganda  : SE quotes : Language Design and Programming Quotes : Random IT-related quotesSomerset Maugham : Marcus Aurelius : Kurt Vonnegut : Eric Hoffer : Winston Churchill : Napoleon Bonaparte : Ambrose BierceBernard Shaw : Mark Twain Quotes

Bulletin:

Vol 25, No.12 (December, 2013) Rational Fools vs. Efficient Crooks The efficient markets hypothesis : Political Skeptic Bulletin, 2013 : Unemployment Bulletin, 2010 :  Vol 23, No.10 (October, 2011) An observation about corporate security departments : Slightly Skeptical Euromaydan Chronicles, June 2014 : Greenspan legacy bulletin, 2008 : Vol 25, No.10 (October, 2013) Cryptolocker Trojan (Win32/Crilock.A) : Vol 25, No.08 (August, 2013) Cloud providers as intelligence collection hubs : Financial Humor Bulletin, 2010 : Inequality Bulletin, 2009 : Financial Humor Bulletin, 2008 : Copyleft Problems Bulletin, 2004 : Financial Humor Bulletin, 2011 : Energy Bulletin, 2010 : Malware Protection Bulletin, 2010 : Vol 26, No.1 (January, 2013) Object-Oriented Cult : Political Skeptic Bulletin, 2011 : Vol 23, No.11 (November, 2011) Softpanorama classification of sysadmin horror stories : Vol 25, No.05 (May, 2013) Corporate bullshit as a communication method  : Vol 25, No.06 (June, 2013) A Note on the Relationship of Brooks Law and Conway Law

History:

Fifty glorious years (1950-2000): the triumph of the US computer engineering : Donald Knuth : TAoCP and its Influence of Computer Science : Richard Stallman : Linus Torvalds  : Larry Wall  : John K. Ousterhout : CTSS : Multix OS Unix History : Unix shell history : VI editor : History of pipes concept : Solaris : MS DOSProgramming Languages History : PL/1 : Simula 67 : C : History of GCC developmentScripting Languages : Perl history   : OS History : Mail : DNS : SSH : CPU Instruction Sets : SPARC systems 1987-2006 : Norton Commander : Norton Utilities : Norton Ghost : Frontpage history : Malware Defense History : GNU Screen : OSS early history

Classic books:

The Peter Principle : Parkinson Law : 1984 : The Mythical Man-MonthHow to Solve It by George Polya : The Art of Computer Programming : The Elements of Programming Style : The Unix Hater’s Handbook : The Jargon file : The True Believer : Programming Pearls : The Good Soldier Svejk : The Power Elite

Most popular humor pages:

Manifest of the Softpanorama IT Slacker Society : Ten Commandments of the IT Slackers Society : Computer Humor Collection : BSD Logo Story : The Cuckoo's Egg : IT Slang : C++ Humor : ARE YOU A BBS ADDICT? : The Perl Purity Test : Object oriented programmers of all nations : Financial Humor : Financial Humor Bulletin, 2008 : Financial Humor Bulletin, 2010 : The Most Comprehensive Collection of Editor-related Humor : Programming Language Humor : Goldman Sachs related humor : Greenspan humor : C Humor : Scripting Humor : Real Programmers Humor : Web Humor : GPL-related Humor : OFM Humor : Politically Incorrect Humor : IDS Humor : "Linux Sucks" Humor : Russian Musical Humor : Best Russian Programmer Humor : Microsoft plans to buy Catholic Church : Richard Stallman Related Humor : Admin Humor : Perl-related Humor : Linus Torvalds Related humor : PseudoScience Related Humor : Networking Humor : Shell Humor : Financial Humor Bulletin, 2011 : Financial Humor Bulletin, 2012 : Financial Humor Bulletin, 2013 : Java Humor : Software Engineering Humor : Sun Solaris Related Humor : Education Humor : IBM Humor : Assembler-related Humor : VIM Humor : Computer Viruses Humor : Bright tomorrow is rescheduled to a day after tomorrow : Classic Computer Humor

The Last but not Least


Copyright © 1996-2016 by Dr. Nikolai Bezroukov. www.softpanorama.org was created as a service to the UN Sustainable Development Networking Programme (SDNP) in the author free time. This document is an industrial compilation designed and created exclusively for educational use and is distributed under the Softpanorama Content License.

The site uses AdSense so you need to be aware of Google privacy policy. You you do not want to be tracked by Google please disable Javascript for this site. This site is perfectly usable without Javascript.

Original materials copyright belong to respective owners. Quotes are made for educational purposes only in compliance with the fair use doctrine.

FAIR USE NOTICE This site contains copyrighted material the use of which has not always been specifically authorized by the copyright owner. We are making such material available to advance understanding of computer science, IT technology, economic, scientific, and social issues. We believe this constitutes a 'fair use' of any such copyrighted material as provided by section 107 of the US Copyright Law according to which such material can be distributed without profit exclusively for research and educational purposes.

This is a Spartan WHYFF (We Help You For Free) site written by people for whom English is not a native language. Grammar and spelling errors should be expected. The site contain some broken links as it develops like a living tree...

You can use PayPal to make a contribution, supporting development of this site and speed up access. In case softpanorama.org is down you can use the at softpanorama.info

Disclaimer:

The statements, views and opinions presented on this web page are those of the author (or referenced source) and are not endorsed by, nor do they necessarily reflect, the opinions of the author present and former employers, SDNP or any other organization the author may be associated with. We do not warrant the correctness of the information provided or its fitness for any purpose.

Last modified: July, 07, 2013