Hello everyone. I need to use voice recognition but have not been able to figure out how to use it sucessfully.

Dragon Naturally Speaking 10 Question:

One of my problems with Dragon seem to be based on an exponential deterioration of voice recognition. The program seems to recognize phrases and commands initially, but as time progresses the program seems to perform worse and worse. Some days it runs poorly from the jump, and others it is better. The program often seems to believe that phrases or words I say are commands, and sometimes it even refrains from responding to the most obvious commands that I give it. For example, When correcting a word you have the option of one of four corrections. I will say "Choose 2", and the program simply does not respond, or it will move and choose another choice altogether. I have increased my ram to 4 GB of DDR2 running at 800mhz, defragged my HDD, ran the program through several training programs, added new words to its recognition list, and I still feel that it is not learning anything that I teach it. Has anyone here had similiar problems? Is there anything that I am personally doing? Is my computer powerful enough to run this program efficiently?


Desktop Specs:

Dell Dimension C521
AMD Athalon 64 Dual-Core Processor 4000+ (2.10 Ghz)
4GB DDR2, 800 mhz
Windows Vista 32-bit

Notebook Specs:

Sony Vaio
Core 2 Duo (1.5Ghz)
4.00 GB DDR2 (800mhz)
Windows Vista 32-bit

bluerhino,

Unfortunately, the one thing that would be helpful in determining why you are not getting good accuracy with DNS 10 is omitted from your post. That is, what microphone/soundcard are you using? Make sure you let us know what your make/model of microphone is as well as the same for your soundcard.

When you run the Audio Setup Wizard volume adjustment where does the blue bar fall (end up) during the microphone volume adjustment phase? Above 75%, below 25%, approximately 50%, what?

When you run the Audio Setup Wizard Quality Check, how high are the yellow bars in the graph why you are dictating the text? How much space is there between the top of the green bars and the top of the yellow bars? To the green bars extend almost to the top of the display area? If not, where do they fall?

When you run the Audio Setup Wizard Quality Check what is the Quality Check Ratio number that you get at the end?

When you complete the Audio Setup Wizard and before exiting the wizard, click the Play button. What is the quality of your voice reply. Is it clear and easily distinguishable (remember that the human ear is 100 times better than the computer at discriminating audio)? Is it muffled? Is there any background noise (i.e., low-frequency hum, crackling or what is normally called white noise [hiss])? Can you hear other background noise (i.e., radio, other people speaking, room sounds, etc.)?

When running the Audio Setup Wizard, Quality Check test portion, jiggle the wire on your microphone (if you are using a wired close talk microphone) and see if it produces any spikes are introduces any crackling noise. Also, do this a couple of times paying particular attention to the connection points to the input Jack and the connection to any mute switch or to the headphones. It is possible that you have a defective microphone, though unlikely, which is shorting out due to broken wires or connections.

Reason that I ask these questions is that the most probable cause for your problem has nothing to do with DNS itself, but has to do with the quality of your microphone/soundcard. That piece of hardware is significantly more critical to accurate speech recognition than any other part of your computer system. Also, if your dictation style, enunciation, pace of speaking, etc. are not natural, this has an impact. Overtraining. That is, you should never have to train more than once, after which you should get at least 98% accuracy. In fact, if you select "None" as the initial general training option when creating a new user, that is the speaker independent Acoustic Model and most users get at least 97% to 98% without ever doing any training whatsoever on that user profile selection. Regardless, if you have to train your user profile more than once, or if you do training more than three times, and you still get poor accuracy, then your Acoustic Model is not optimal. Running multiple training sessions is not recommended because on a poor initial Acoustic Model, multiple trainings tend to degrade user accuracy anyway.

Keep in mind that your experience relative to the accuracy that you are obtaining from DNS 10 is the extreme exception rather than the rule. Therefore, there is a reason for this and it is most likely in your microphone/soundcard or the manner in which you dictate, or both.

I await your responses to my questions, after which you will take it to the next level.

Chuck Runquist
Owner, GEMCCON - The Choice of Intelligence
Speech Recognition Consulting and Training

If there's more than one way to do a job, and one of those ways will result in disaster, then somebody will do it that way. (variant of Murphy's law - Edward A. Murphy, Jr.)

Good Afternoon,

Thank you for you prompt response. Hopefully, my answer will be as insightful and informative as your initial post.

My Sound equipment setup is the following:

Microphone: Logitech Orbit (Desktop mic) -- this is a USB based microphone.
Sound Card: SigmaTel High Definition Audio

The Microphone volume level is around 33%
When using the Bar graph during the Audio Quality check, the low-end of the spectrum remains relatively high (without going off the chart I believe), or at least higher than the rest of the graph. The yellow bars are around half the height of their corresponding green bars.

On my Desktop when testing the Speech-to-noise- Ratio, I received a 15.
The audio playback does seem to be muddled, particularly on the high-end.
My assistant has recommended possibly using a condensor mic for clarity, but also suggest that perhaps this is going to an extreme setup due to the requirement of a preamp, etc.
He also questioned the soundcard and suggested that perhaps this isn't a high-end card, but something Dell threw in to save money.
When we ran the audio quality check and he jiggled the wire, it was hard to see any spikes in the graph, possibly due to the readout not being in real-time.
Although, there appeared to be no spike in the audio.

When I tried to run the Acoustic and Language Model Optimizer I was told there was an insufficient amount of new data in order for it to run.


As far as the notebook, I am using a built-in microphone (located on the monitor -- I believe).

I guess Chuck is away. Any other people have ideas?

Chuck probably nailed the problem. Your microphones aren't suited for speech recognition.
I've got some of my thoughts on microphones at http://www.pcspeak.com/products/microphones.shtml

Other speech recognition dealers will have somewhat different recommendations. www.microphones.com, www.emicrophones.com, www.speechrecsolutions.com, www.knowbrainer.com all have material worth reading about microphones for speech recognition.

Going to a condenser microphone would be better than what you have now, but while they will work for speech recognition, they aren't ideal for it.

Larry Allen
http://www.pcspeak.com

Unfortunately you almost couldn't have made a worse microphone decision. The Logitech Orbit microphone is fine for VoIP and gaming but it is completely unacceptable for speech recognition. Speech recognition requires a premium microphone. Furthermore, very few conference style (distance) microphones work with NaturallySpeaking. You should typically be within less than 2 inches of the microphone element. We recommend contacting a professional speech recognition vendor or minimally checking out our Microphone Comparison Matrix where you can compare pricing, comfort, accuracy, noise cancellation and features. If you like the basic overall design and style of working without a headset microphone, you might want to check out the Olympus ME 31 gun microphone which is one of the few microphones that allows you to dictate from a distance.

Thank you again. At one point, I had compared using a head set as opposed to a microphone that you place on a table and it seemed at that time that there was no difference. Obviously, I was wrong. I already have Sennheiser PC 166 USB Stereo Multimedia Gaming Headset which I used briefly before making that decision. I have now plugged that in and have started using it. I do note a significant improvement. Over the next few months, I will probably buy the most appropriate headset, but I think that this will give me some improvements right away.

When I switched the microphone system this morning, Dragon informed me that because of system resources, it will disable certain functions including the ability to use the Dragon voice recognition program with Microsoft Word. Since a friend of mine increased the memory from two RAM to four RAM, the guilty party has to be the processor. I probably need to buy a new computer this coming fall as well.

I will look through the websites that you two gentlemen have suggested. I just wanted to let you know that there has been a fairly dramatic improvement as a result of switching microphones. It was so simple but I never would have thought of it on my own because I thought that I had already tested that element.

The Sennheiser PC 166 USB microphone isn't the best Sennheiser microphone for use with NaturallySpeaking but it is adequate.



If NaturallySpeaking thinks that your processor isn't up to the task or you don't have enough RAM, NaturallySpeaking can automatically disable some functionality but if you wish, you can optionally return that functionality by clicking DNS Options/Commands/Natural Language Commands and placing a checkmark in Enable Natural Language Commands.

I'm having the same problem with DNS 10, but my audio setup is perfect. Recordings using my mic sound crystal clear. Speech-to-noise ratio is 23. Any idea what's causing this?

This may be pure coincidence, but one day I was practicing my guitar for an hour with the computer microphone on. After my practice hour session, Dragon would not recognize anything I was saying! I had to create a new user profile and start fresh. Since then it has been getting better every day ( i only bought it ten days ago ). So now I always say 'go to sleep' when not using it so it doesn't catch stray sounds. What do you think?

I think you're doing the right thing! You should always use 'Go To Sleep'

If you're not going to use your computer for a while it is preferable to turn the microphone completely off rather than just putting it in Sleep Mode. When you put your computer in sleep mode it may accidentally wake up when it misinterprets a phrase because NaturallySpeaking is still listening when it's in sleep mode. You can also overflow the buffer when the microphone is in Sleep Mode although Ver. 10 usually handles buffer overrun problems well. Bottom line: If you're not going to be using your computer for more than 20 minutes, you should turn your microphone completely off rather than placing it in Sleep Mode.

Thank you again. Using the new headphone headset, I did the audio quality test and scored 24 instead of 15. This is a significant improvement. It said that I passed, which I think is what it said the last time with the other microphone. The program probably should have warned me with the Logitech microphone that even though 15 was passing, it was only minimally acceptable and that I should consider getting a better microphone. Anyway, there is a noticeable improvements now. I'm actually using voice-recognition to dictate this message. What would be useful is to know -- what is the noise ratio scale? What is the highest number that you should get or the most desired number?

I would be interested in upgrading at some point. I get the impression that some of you gentlemen are professionals in this field and so I would certainly prefer to buy from one of you, if that is possible. I would think that a better headset would run between $150 and $250. I would be open to suggestions. I would probably upgrade in the fall.

Now that I have this particular issue under control at least temporarily, I will ask my friend to assist me with some of the commands. I have struggled terribly over the past couple of years, but I have to see how much of it will go away now that I have a better microphone. I will then need to figure out how to perfect the way that I make commands in order to prevent it from getting confused with regular dictation.

You have all been wonderful. Thank you again.

(Yes, Lunis is right about turning the mike off rather than relying on 'go to sleep'...)

I get the impression that some of you gentlemen are professionals in this field and




What would be useful is to know -- what is the noise ratio scale? What is the highest number that you should get or the most desired number?

There isn't a highest number you should get, there's no one-to-one relationship between score and recognition, and some very good mikes score relatively low. Judge by recognition rather than score. (Having said that, all else being equal, a higher score is better.)

I am really behind with mikes so won't recommend one.

I have struggled terribly over the past couple of years, but I have to see how much of it will go away now that I have a better microphone.

Yes -- sorry you've struggled -- you may also need to look into training technique and so on. But things should start to be better now.