First:  I would like to wish to all of you the appropriate seasonal solstice greetings and wish for all of us worry free dictation for 2012. :-)

 

Until sometime ago, I was using my Sony SX fifty-seven recorder quite happily. One of the peculiarities I had noticed was that the little Sony was more accurate as a microphone than any other device I have tried.

 

Having spent some several months (actually almost a year now realise) fighting Server problems, I have endeavoured to get back to serious volume writing. Something seems to have happened to seriously impair the performance of my system in transcribing material from the little Sony. I switched to my Philips backup unit and experienced exactly the same problems:

Not only has the accuracy suffered greatly, but the actual transcription stops dead approximately halfway through the transfer.

I am about to try and enroll a new user, but thought I would possibly find some very bored member of the "club" out at sea or in the army who just might be able to help . I have also just noticed that in doing this small posting I have lost several single syllable small words from my dictation strings.

 

Oh, another small query:

Dragon is adamant that you do not do anything else while transcribing and especially not move your cursor and I'm not even sure if we are not allowed to go for coffee (I have done so and bumped my desk inadvertently and been punished for it) and I am wondering if in this day and age of fairly powerful units, if it is really necessary and if not is it avoidable in some tricky way? :-)

Cheers!

Himagain

-------------------------
2010 Samsung R780 Luggable. 4Gig , 1 terabyte Drive, W7 +   Ubuntu 10+,  Sony SX57 Recorder, Philips LFH617 Digital Voice Tracer + Dragon 11.5 Preferred

The average condenser microphone cost $2.50 to manufacture and is surprisingly accurate. The disadvantage of condenser microphones is that they are somewhat easy to overload but this isn't likely to be a problem for dictation. However, when you hold a microphone/recorder in your hand for dictation purposes, the distance tends to vary and this can have some effect on your dictation. We are not surprised that your Sony SX 57 is more accurate as a microphone then as a transcribing device but the good news is that it can double as a USB microphone on steroids. Unfortunately, neither the Philips LFH617 or Sony SX57 digital recorder algorithms are well suited to transcription. That's why neither unit made our DNS digital recorder recommendation list.

Quote:

One of the peculiarities I had noticed was that the little Sony was more accurate as a microphone than any other device I have tried.

What other microphones have you tried and perhaps more importantly, were they USB microphones. 1 of the advantages of using your digital recorder as a microphone is that it is bypassing your integrated sound card. Normally, a good microphone will perform a little better than a digital recorder but you have another option: If you have a good analog microphone, you could optionally plug it into your digital recorder which would not only give you the advantage of turning your microphone into a USB microphone but additionally give you the advantage of auto level control which is only available in a handful of professional microphones and could boost your accuracy.

Are you, by any chance, plugging your digital recorder into a USB hub or front or side USB port on your computer (assuming this is a desktop computer)? These ports are fine for transcribing because they're only used for downloading data and the occasional timeout won't be a problem but you shouldn't use a nondedicated port for direct microphone input. Note that all notebook USB ports are direct so if you're using a notebook computer, you can ignore the previous advice.

Considering that you're experiencing this issue with both of your recorders, the USB port sounds like the most likely culprit. If you are already using a dedicated USB port, consider switching to a different dedicated port.



NaturallySpeaking has to seize total control of your computer in order to transcribe digital recordings. We currently know of no workaround but depending upon the speed of your computer and the quality of your recordings, NaturallySpeaking should be able to transcribe your dictation at approximately 20 times the speed it was dictated (seemingly roughly the speed of light). The cleaner your recordings are, the faster and more accurately NaturallySpeaking can work. If your digital recordings are transcribing in real-time or even triple time, NaturallySpeaking is probably struggling which will result in lower accuracy and longer transcription times. In other words trash in and trash out.

-------------------------

Click KB 2012 REV D to Download a 30 Day Evaluation of KnowBrainer 2012 

 

 

 

Hello Lunis,

Thank you for such a (typical) comprehensive reply especially at this time! :-)

The key part of your response was the statement that transcription should be 20 times the dictation rate! 

It has always been pretty pathetic and done in "down times" like dinner for its painfully slow speed.

This has been my experience through multiple evolutions with Dragon and my various systems.

I'm now using a Portable Samsung R780, so the port choice is ok, unless it is a dud??? Will change it around.  My daily standby unit is a Logitech desktop unit set for rapid sleep and it works reasonably well for its type and my environment. I use it on basically near fastest Dragon setting and it sure beats my typing accuracy, much less speed! :-)

I've changed my methodology of writing to doing a multi-tasking event of dictation while walking on my (near) private beach in the mornings, so some losses must be expected! But except for seagulls and crows (which add rude words to the text), it copes quite well with the sound of the sea. I do use a cloth mask around the unit usually, but often forget.

However, if I can resolve this painfully slow (less than my dictation speed ) transfer rate, this will be a good Xmas present!

Thanks again,

Himagain 

 

 

-------------------------
2010 Samsung R780 Luggable. 4Gig , 1 terabyte Drive, W7 +   Ubuntu 10+,  Sony SX57 Recorder, Philips LFH617 Digital Voice Tracer + Dragon 11.5 Preferred

Well, 20 times the dictation may be saying a bit too much. A while back, we did some extended performance testing with different computers (different in speed) being involved, and we were using digital recordings in order to keep the input constant. The best (fastest) machines achieved a performance factor of about 0.2, meaning, it took about 2 seconds to transcribe an utterance of about 10 seconds. So make that 5 times faster (on a very strong computer) to about 3 times faster (on an average machine).

Rüdiger

 

-------------------------

Well, it's past the point where we can make any changes in the code, but we can still make changes to the Easter Egg!

 

 Happy New Year Rutiger,

 There is nothing like having facts and statistics to help you arrive at certain conclusions in life. While your test figures are nowhere near as exciting to hear about as Lunis' comment :-) they certainly point to something substantially at fault with my results from using Dragon 11.5.

My own test conditions are more approximating my real-world and one of the test dictations for example, (a favourite) is Dave Barry's contribution (about 20 min duration). I was hopeful that I would be able to compare results from my outside recording exercise with as near as I could get to good conditions to do it in, at home.

Not only is the transcription painfully slow, definitely far slower than my dictation speed (because my clever little Sony actually wipes out long pauses all on its own) but I have now realised that it quits after perhaps five to seven minutes.

 Something is severely wrong with my system. Even dictating here, if a typical error takes place, the correction side is very difficult as for instance, after the correction, spacing will be incorrect on those times when I can select a replacement option as distinct from over typing.

I cannot make a simple case change inside this forum for instance. The voice instruction to capitalise a preselected letter simply doesn't take place.

then, to do this simple exercise add-a-line, it does not capitalise the first letter despite it being a follow on from a full-stop. 

 Another strange anomaly with my Dragon system -  recently discovered, is that I cannot train a new voice via a recorder. Nothing I do will make Dragon find the new Sample voice dictation.

Perhaps it is time for a reinstall?

 Cheers?

Himagain

-------------------------
2010 Samsung R780 Luggable. 4Gig , 1 terabyte Drive, W7 +   Ubuntu 10+,  Sony SX57 Recorder, Philips LFH617 Digital Voice Tracer + Dragon 11.5 Preferred

Select-&-Say (Full Text Control) should be available on this forum if you're using Internet Explorer (not Firefox or Safari) and you remember to launch NaturallySpeaking before other Full Text Control applications; especially Microsoft Word, Outlook and WordPerfect.

It's also possible (think of this as Plan B) to force the 1st word of a new line to appear in Title Case even if there is no period at the end of the preceding line. Just change the properties of the Format the next word in the Vocabulary Editor for the new line command. The following is an example of this change being applied to the next line command which is the same as new line.

                


-------------------------

Click KB 2012 REV D to Download a 30 Day Evaluation of KnowBrainer 2012 

 

 

 

Initially, the thought of you wearing five finger shoes and a gangster hat on the show seemed really cool to me, but wouldn't that be overdoing it somehow?

Happy New Year!

Rüdiger

 

-------------------------

Well, it's past the point where we can make any changes in the code, but we can still make changes to the Easter Egg!

Quote:

but wouldn't that be overdoing it somehow?

Only if you are pretending to be eccentric

-------------------------

Click KB 2012 REV D to Download a 30 Day Evaluation of KnowBrainer 2012 

 

 

 

Hi Lunis,

I sure live and learn!

THIS explains much:

Select-&-Say (Full Text Control) should be available on this forum if you're using Internet Explorer (not Firefox or Safari) and you remember to launch NaturallySpeaking before other Full Text Control applications; especially Microsoft Word, Outlook and WordPerfect.

As I don't know anyone who would still use Internet Explorer, much less a Security Fanatic such as myself, this would never have occurred to me. I do avoid using Microsoft applications for many reasons so I was pleased to see Your alternative approach and did go and adjust my settings in Dragon. Unfortunately, it didn't fully work. This is being dictated using Firefox and with no other text type applications open.

 Because of my workload, I customarily would have several applications open at any given time (even different types of text capture programs) and I cannot envisage using Microsoft Explorer as I am continually visiting many different websites.

Luckily, with the exception of KnowBrainer forums and one or two other extremely valuable reference/support sites, I can probably survive by hand chiselling text! :-)

Or, put up with the extremely difficult correction process shortfall of Dragon itself.

 

As usual, thank you again for revealing the occult mysteries that lay behind the well named Dragon of  Chinese mythology! :-)

Himagain

-------------------------
2010 Samsung R780 Luggable. 4Gig , 1 terabyte Drive, W7 +   Ubuntu 10+,  Sony SX57 Recorder, Philips LFH617 Digital Voice Tracer + Dragon 11.5 Preferred

Qualitative researchers have evolved their methods continually, often due to technological breakthroughs that have enabled them to collect, analyze or present data in novel ways or to obtain a stronger authenticity or reflection of participant perspectives.

-------------------------
Download Frostwire