Chuck Runquist
Technical Project Manager
VoiceTeach LLC
Home of VoicePower® Ultimate

"Kindness is the language which the deaf can hear and the blind can see." -- Mark Twain

Any feedback or direction anyone can provide would be greatly appreciated.

Chuck Runquist
Technical Project Manager
VoiceTeach LLC
Home of VoicePower® Ultimate

"An error does not become truth by reason of multiplied propagation, nor does truth become error because nobody sees it" -- Gandhi

 I don't know if things have changed.  

 An alternative is to get a USB, headset mike or whatever your favorite input device is and forgo an outboard USB pod.  

Just barely.  Consider the following:

1.  Your laptop only supports 32-bit applications.  It does not support 64-bit.  Therefore, while you may be able to run Windows Vista or Windows 7 32-bit, and that's questionable, because your CPU may be one of those that sits on the cusp and will not support the newer versions of Windows.  Therefore, you may be stuck with Windows XP.

2.  If you haven't currently updated Windows XP to SP3, and you're still currently running Windows XP SP2, you can run into some problems: (1) Microsoft is no longer supporting Windows XP SP2 as of July 1 of this year, (2) some current applications, including DNS, require MSXML 4.0 or higher and Microsoft’s NetFramework 3.51 SP1 or higher, (3) unless you have consistently accepted and applied the Windows updates, you may be missing you may be missing critical updates which are required for some of the newer applications and even for some of the current application base.  In addition, if you aren't currently updated to Windows XP SP3, your system is vulnerable because Microsoft is no longer supporting security updates for SP2 and some of the updates that may be required are specifically related to SP3 and wouldn't be available even if you maintained your system with regular updates to SP2 all the way up to July 1, 2010.

3.  The processor (CPU) in your system, while it is a Core2™ dual core processor, only contains 2 MB of L2 cache (1 MB dedicated per core).  This can significantly impact on performance because the newer Core™ i7 and even the more recent but older Core2™ Duo and Core2™ Quad processors have significantly more L2 cache and the Core™ i7 processors support L3 cache (SmartCache) which is shared across all processors (something that L2 cache does not do and which also affects the performance of DNS, particularly DNS 11) .  The reason that the cache affects performance is that caches magnitude of times faster than RAM.  If the information that is being sought is stored in the cache, and cash is constantly updated, then the time required to go from dictation to display of text (speech to text) will be fairly quick.  However, the smaller the cache, the less frequently used information can be stored in it, which necessitates then that the processor must go out to the RAM or to the hard drive in order to find the information that it needs to process your speech.  This is magnitudes of times slower.  In general, the ideal cache size is approximately somewhere between 4 and 6 MB.  Since the cash is always being updated, the smaller the cash the more often the information contained in such is altered.  What the processor might be looking for relative to executing recognition of your speech is less likely to be in the cash the smaller it is.  In addition, SmartCache, which is cash that can be shared across all the cores in a particular processor, enhances DNS performance because it is more likely that the information contained in the L3 cache will be current, and no matter what core is being used to process your speech, that core has access to all of the cache (L3 cache), not just its own dedicated cache (L2 cache).  So, with L3 cache (SmartCache) the likelihood that the information being sought by the processor relative to processing your speech is more likely to be there.

4.  The total amount of RAM on your system is more than adequate, even though the RAM speed might be slightly slower than current DDR3 RAM.  However, the processor speed is no longer as critical to DNS performance as is RAM and cache.

So, the bottom line is that your laptop meets basic and minimum requirements for using DNS 11, but it won't blow the doors off of Ferrari it comes to speed.  Also, depending upon how DNS 11 analyzes your hardware during installation and the creation of a user profile, it's possible that your system might support BestMatch IV.  That is it might attempt to use the multicore/multithreading capability of DNS 11.  From a pure accuracy standpoint there is no basic difference between BestMatch IV and BestMatch III.  The entire BestMatch algorithm has been improved in DNS 11 and the only difference between BestMatch IV and BestMatch III is the fact that BestMatch IV uses two cores and by virtue of creating two Acoustic Models, which it uses to process your dictation by making a pass through each Acoustic Model and then merges the results picking the nine best matches and displaying the one with the highest priority (#1).  However, this process is memory and resource intensive and greatly slows down the process on older processors.  In this case, your RAM is fine, if you're processor that may slow things to a crawl between your dictation and the display of the text in your document.  Therefore, if installing DNS 11 selection BestMatch IV and improves exceedingly slow for you, there are couple things that you can do:

1.  Create a new user profile and manually select BestMatch III via the Advanced button when selecting your vocabulary in the new user dialog.

2.  You can try adjusting the Speed vs. Accuracy slider, but don't go below the 50% default setting.  The reason for this is that as you move the Speed vs. Accuracy slider below 50%, you lose the ability to use the four word context n-gram model (quadgram model).  This has no effect if you're dictating in short choppy phrases of three words or less, but it can impact accuracy if you are dictating in longer phrases (utterances) of four words or more because below 50%, you're only using the to word and three word context models (bigram and trigram models).  When you're not using the quadgram model, it becomes more difficult for DNS to predict contexts, and thus decreases your overall accuracy.

The bottom line is that DNS 11 will work for you, but probably not ideally.

Chuck Runquist
Technical Project Manager
VoiceTeach LLC
Home of VoicePower® Ultimate

"About the time we can make the ends meet, somebody moves the ends." Herbert Hoover

Thanks,

muckworm

Thnks,

Matthew

http://www.talk-tech.net/pages/sylencer.html

I hate it because, well... it's hard enough to breathe properly while using speech recognition software. Why would you want to wear something that inhibits your breathing? That said, there are circumstances where such a setup is practical and productive. This might be one of them.

Thank you for any info you may have! 

At the risk of bothering you guys further, I have one remaining point I need clarification on.  I've somehow gotten the idea a usb sound pod is essential to maximizing any microphone's performance with speech recognition.  Is that a mis-conception on my part? Would I receive satisfactory results plugging the ME-31 or Q7 directly into my HP Pavilion and depending on the onboard sound? I'm thinking the usb pod would help in any situation but am curiouis what the "bang for the buck" return is on adding it to my purchase of either the Shure or the Samson? 

Thanks, again for devoting the time and energy to my questions here. I appreciate you both. 

PS:  Just noticed that the Samson connects via XLR so my question about connecting directly to the PC doesn't apply.  Looks like I'll have to find some sort of XLR-to-DAC adaptor to be able to use this microphone.

Just my $.02.

Chuck Runquist
Technical Project Manager
VoiceTeach LLC
Home of VoicePower® Ultimate

Education is when you read the fine print. Experience is what you get if you don't. - Pete Seeger