Blue Tooth Microphones for Speech Recognition

Our company policy in the past has been to not post results of microphones that we are in the process of testing or ones that have failed to meet our standards. Please see the link on our home page for How We Test Our Microphones. The past six months we, along with Graham Hendry of IT Speaking in the UK, have been working with a well-respected, and well-known manufacturer of speech recognition microphones to refine for speech recognition purposes, their cell phone Bluetooth wireless unit.

As has been apparent the past two years, creating a Bluetooth wireless device that will work well with speech recognition is not a simple matter. The first problem is the actual signal itself. It is common for the Bluetooth signal to have dropouts which of course affect recognition accuracy of not only the missing words but possibly also the context clues of the words before and after. The words before and after are essential for speech recognition software to be accurate. The company we are working with seems to have fixed this problem in the last version tested.

Other things that we feel are important, and lacking in other Bluetooth microphones not specifically designed for speech recognition, include the length and design of the boom. This is needed in order to keep the microphone element in the proper placement off the corner of one's mouth. A noise canceling microphone element is also essential. The company we are working with has listened and we expect to test the preproduction unit in the next several weeks.

The other modifications will allow for the base station to connect to the PC through the soundcard or with a supplied USB full duplex pod. This will allow the best transfer of audio signal directly from base station to soundcard/USB audio drivers. This is much better than dealing with after-market USB dongles that provide unreliable audio transfer.

Perhaps the best news will be the price. The company developing and manufacturing this device promises that the price will be competitive or lower than any other similar Bluetooth device out there. As soon as Marty and I receive this we will thoroughly test and hopefully be offering it within the next couple of months.

Michael Mendick,
eMicrophones, Inc.
http://www.emicrophones.com

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Any progress?

Michael,

I'm waiting for further comments from you before investing in a Bluetooth microphone. Any progress on the project you described above?

Charles

Any Progress

Charles
(Not Michael) We are on to the next phase of testing - so far so good.

Graham Hendry

What's the frequency range

What's the frequency range of this bluetooth microphone?
Thanks,
Joey

I believe this microphone

I believe this microphone range has finally been released along with one or two accessories.

I hope our resident gurus can provide some more details for the eagerly awaiting crowd.

Graham

According to

According to emicrophoneswebsite BTH (Bluetooth headsets) day is tomorrow. Will it happen?
Quentin

In the meantime check out

In the meantime check out this thread:

http://forum.knowbrainer.com/read.php?f=2&i=24745&...

Graham

emiliaserv wrote:In the

emiliaserv wrote:

In the meantime check out this thread:

http://forum.knowbrainer.com/read.php?f=2&i=24745&t=24745

In that posting, I note that there is a reference to the six hour continuous use lifetime of the battery.

If the Bluetooth used rechargeable batteries, it would simplify matters, since two or more sets of batteries, there would always be recharged ones available. The writer of that article was talking about getting two receivers, but it would be much cheaper to have rechargeable batteries.

Therefore my simple question is – can they take rechargeable batteries?

I think you will find they

I think you will find they are rechargeable batteries, the person in question wanted to buy a spare unit so that he would be able to use the device even when the first set was being charged, not sure why he couldn't just buy a spare set of batteries. At least that is how I understand it but I may have got hold of the wrong end of the stick.

Graham

emiliaserv wrote:I think you

emiliaserv wrote:

I think you will find they are rechargeable batteries, the person in question wanted to buy a spare unit so that he would be able to use the device even when the first set was being charged, not sure why he couldn't just buy a spare set of batteries. At least that is how I understand it but I may have got hold of the wrong end of the stick.

The batteries are rechargeable but not replaceable. They remain inside the headset. This makes sense because replaceable batteries would be much heavier and larger to give six hours talk time.

VXI Bluetooth headsets have been available for several weeks as headset and charger for use with computers that are already Bluetooth enabled or with the use of a USB dongle it will enable any computer as Bluetooth.
See:
http://www.emicrophones.com/microphones/product_de...

The version that comes as a base station that works as a receiver and charging unit is still approximately two weeks away.

--
Martin Markoe, eMicrophones, Inc.
The best microphones for Speech Recognition
See us at: http://www.eMicrophones.com/index.asp
Read, "Key Steps to High Speech Recognition Accuracy" at:
http://www.emicrophones.com/docDetails.asp?Documen...

Martin Markoe wrote:The

Martin Markoe wrote:

The batteries are rechargeable but not replaceable. They remain inside the headset.

Being pedantic the batteries are replaceable (in the event of failure) but only with difficulty and a screwdriver. Smiling

BlueParrott TalkPro hadsets and charger have also been available in the UK for the past week.
See:
http://www.itspeaking.co.uk/speech-producttype.asp...

Graham (Hendry)

For those of us who may be

For those of us who may be technologically (and perhaps financially) challenged, exactly what are the advantages of these new SR-enabled Bluetooth microphone sets? Are they better than, say, current wired mikes, or are they designed for certain types of settings?

Bruce

PS: Maybe an early adopter or two would be motivated to post a review in the "product reviews" section of this site.

Bruce, > what are the

Bruce,

> what are the advantages of these new SR-enabled Bluetooth microphone sets?

The idea of Bluetooth is that the receiver portion is built into the computer. In most radio frequency wireless units the receiver is a large unit that requires connection to a wall outlet. Within a couple of years you will probably fine Bluetooth built into every computer. It's currently built into most new high-end notebooks. As a matter of fact, it's very simple to make any computer a Bluetooth enabled computer by adding a USB dongle. These are smaller in size of the microphone and connect to the USB port.

Bluetooth devices are not limited to microphones. There are currently dozens including keyboards and mice.

--
Martin Markoe, eMicrophones, Inc.
The best microphones for Speech Recognition
See us at: http://www.eMicrophones.com/index.asp
Read, "Key Steps to High Speech Recognition Accuracy" at:
http://www.emicrophones.com/docDetails.asp?Documen...

Martin Markoe wrote:Bruce,>

Martin Markoe wrote:

Bruce,

> what are the advantages of these new SR-enabled Bluetooth microphone sets?

The idea of Bluetooth is that the receiver portion is built into the computer. In most radio frequency wireless units the receiver is a large unit that requires connection to a wall outlet. Within a couple of years you will probably fine Bluetooth built into every computer. It's currently built into most new high-end notebooks. As a matter of fact, it's very simple to make any computer a Bluetooth enabled computer by adding a USB dongle. These are smaller in size of the microphone and connect to the USB port.
**********************************************************************************************************

The following is written in the second person since I had initially intended to be an e-mail message to Marty Markoe. However, I decided to post it here for comments from the group.

Marty,

A little over a month ago I purchased the BlueParrot VXI B1000 GTX and the VXI B50 Charger Unit from you. I had hoped to be able to use the headset alone with my Compaq notebook computer, that has built in Bluetooth connectivity, while leaving the USB base station permanently connected to my desktop computer. However, the notebook computer does not seem to be able to find and pair with the headset even though I have installed proprietary Bluetooth drivers from the computer manufacturer's web site.

I am forced to resort to the cumbersome procedure of disconnecting and reconnecting the USB cord and the power supply when moving the base station back and forth between the two computers at home, and taking the base station with me when I travel with the notebook.

I realize that you forewarned me about the potential for this problem, and that you volunteered to take the merchant's eyes (I left this misrecognition in for your amusement) merchandise back without a restocking charge if the problem could not be resolved. However, I chose not to take advantage of your offer because I wanted to have a wireless headset available for use with my desktop.

It seems, therefore, that at this point, the only viable option is for me to purchase a dongle for use with the notebook. I note that you are now advertising a cheaper version of the dongle than the XoVox unit recommended heretofore. I assume you have found that this (nameless?) one functions equally well.

One other disappointment -- I find that I cannot leave the headset turned on for more than three hours without it needing to be recharged. To go from fully discharged to fully charged requires 12 hours. This was a surprise to me since, prior to deciding to purchase, I had had read the following comment by Joe Graney on the on the KnowBrainer Forum:

" The battery life is 6 hours, but it seems that the battery is only draining during actual talk time; it appears to go back into standby mode when my Dragon microphone is off. Thus the microphone is lasting me all day long (10 hour day in the office) on one charge."

Speechcontrol.com. advertises: " Standby time: up to 100 hours / Talk time: up to 6 hours". In my mind, using the phrase "up to" in conjunction with a number, makes the alleged information ambiguous enough to be rather meaningless.

I could find nothing on the VXI website to indicate what my expectations should be in this regard.

Charles

Charles Devonshire wrote: I

Charles Devonshire wrote:

I realize that you forewarned me about the potential for this problem, and that you volunteered to take the merchant's eyes (I left this misrecognition in for your amusement) merchandise back without a restocking charge if the problem could not be resolved. However, I chose not to take advantage of your offer because I wanted to have a wireless headset available for use with my desktop.

It seems, therefore, that at this point, the only viable option is for me to purchase a dongle for use with the notebook. I note that you are now advertising a cheaper version of the dongle than the XoVox unit recommended heretofore. I assume you have found that this (nameless?) one functions equally well.

Forewarned is forearmed. Compaq, HP, Acer and Dell Bluetooth have been problematical. Most people who have downloaded the newest Bluetooth drivers from these Manufacturers have been successful. The new drivers have to be forced by going to Device Manager, uninstalling old drivers and installing new drivers. It sometimes takes more than one try and a reboot to do this. VXI has the latest Bluetooth, but the aforementioned computers did not.

The AnyCom Bluetooth USB dongle was found in our search for a less expensive thumb sized Bluetooth USB radio. After testing the AnyCom, to our surprise, when the XoVox dongle label was removed it said AnyCom. The AnyCom we supply is not only $40.00 less expensive than the XoVox, but has the 100 meter (300') radio rather than the 30 meter. Performance of the dongles is identical. The XoVox uses a different software though it is no better than the software supplied with the AnyCom.

As for battery life, are you pushing the multi-function button on the headset to put it into standby mode when not dictating? If not, that is what needs to be done. If you still get shortened battery life, contact eMicrophones directly through our easy to use Contact Us button. This sends the E-mail message to my partner Michael, shipping manager Matthew and to me. We will arrange for VXI to fix it.

--
Martin Markoe, eMicrophones, Inc.
The best microphones for Speech Recognition
See us at: http://www.eMicrophones.com/index.asp
Read, "Key Steps to High Speech Recognition Accuracy" at:
http://www.emicrophones.com/docDetails.asp?Documen...

Apart from also functioning

Apart from also functioning as a charging unit, does the base station offer any advantage over the dongle?

Charles

Charles Devonshire

Charles Devonshire wrote:

Apart from also functioning as a charging unit, does the base station offer any advantage over the dongle?
Charles

The base unit of the VXI Bluetooth has a more powerful antenna. It works through several walls and up to 100' in my home office environment.

--
Martin Markoe, eMicrophones, Inc.
The best microphones for Speech Recognition
See us at: http://www.eMicrophones.com/index.asp
Read, "Key Steps to High Speech Recognition Accuracy" at:
http://www.emicrophones.com/docDetails.asp?Documen...

joey wrote:What's the

joey wrote:

What's the frequency range of this bluetooth microphone?
Thanks,
Joey

This question was never answered. Is frequency response an important consideration in deciding what microphone to purchase? If it is not significant now, might it not be so in the future with the advent of Windows Vista?

Charles

Charles Devonshire

Charles Devonshire wrote:
joey wrote:

Is frequency response an important consideration in deciding what microphone to purchase? If it is not significant now, might it not be so in the future with the advent of Windows Vista?
Charles

A microphone, whether wired or wireless, is an audio input device. If it works with Windows, there is no reason it should not work with any speech recognition system.

Bluetooth wireless, I believe, delivers a signal at 8 kHz. Microsoft Speech uses 16 kHz. For the VXI Bluetooth it does not seem to matter. Performance in Dragon and Microsoft Speech is equally good.

--
Martin Markoe, eMicrophones, Inc.
The best microphones for Speech Recognition
See us at: http://www.eMicrophones.com/index.asp
Read, "Key Steps to High Speech Recognition Accuracy" at:
http://www.emicrophones.com/docDetails.asp?Documen...

Marty, As you know, based on

Marty,

As you know, based on answers given to the questions above,I've decided to purchase the VXI Bluetooth system. In our phone conversation you were kind enough to give me advance warning that I might experience compatibility problems between the VXI and my Compaq notebook. I believe you said that this was based upon feedback from one of your customers who had trouble getting the built-in Bluetooth receiver on his Compaq notebook to recognize the VXI microphone and/or base station.

I wonder if you would do me the favor of putting me in touch with that person by giving him my e-mail address and/or telephone number and asking him to contact me. He may have already resolved the problem. I will be traveling with the notebook the week after Christmas and would very much like to use the microphone and show it off to my friends and relatives during that time.

I have posted the specs for my computers on my blog in case someone else might have some input.

Thanks,

Charles

The impression I get from

The impression I get from some reviews on the web is that the Plantronics wireless headset which uses DECT technology is better in terms of accuracy than bluetooth microphones. Has anyone used this before. I found one on ebay.co.uk for around £165 and I will like to check people's comments before taking the plunge

builder wrote:The impression

builder wrote:

The impression I get from some reviews on the web is that the Plantronics wireless headset which uses DECT technology is better in terms of accuracy than bluetooth microphones. Has anyone used this before. I found one on ebay.co.uk for around £165 and I will like to check people's comments before taking the plunge

I tested it and found noise cancelling ability to be poor. Graham Hendry of www.itspeaking.com is a UK speech recognition dealer with many years experience. He had the same evaluation.

There is hype and then there are people in this business who actually test highly hyped devices.

--
Martin Markoe, eMicrophones, Inc.
The best microphones for Speech Recognition
See us at: http://www.eMicrophones.com/index.asp
Read, "Key Steps to High Speech Recognition Accuracy" at:
http://www.emicrophones.com/docDetails.asp?Documen...

But this guy from 21st

But this guy from 21st century eloquence (another expert) pays glowing tributes to this headset, and Sacnsoft/Nuance is bundling it with Daragon naturally speaking 8. Check here :
http://www.voicerecognition.com/board/index.php?sh...

He says " ThePlantronics CS-50 USB Wireless Headset Microphone was tested with Dragon NaturallySpeaking v8 Medical. The results were as follows.

Speech to Noise ratio: 28

The initial dictation consisted of 669 words, there were 7 mistakes found, which calculates to 99.0% accuracy on the first round of dictation."

I am tempted to buy this , but will like more imput from users here.
Builder

builder wrote:But this guy

builder wrote:

But this guy from 21st century eloquence (another expert) pays glowing tributes to this headset, and Sacnsoft/Nuance is bundling it with Daragon naturally speaking 8. Check here :
http://www.voicerecognition.com/board/index.php?sh...

He says " ThePlantronics CS-50 USB Wireless Headset Microphone was tested with Dragon NaturallySpeaking v8 Medical. The results were as follows.
Speech to Noise ratio: 28

The initial dictation consisted of 669 words, there were 7 mistakes found, which calculates to 99.0% accuracy on the first round of dictation."

Experienced users of speech recognition software like Dragon NaturallySpeaking, understand the audio setup wizard signal to noise ratio is tuned to give good results with inexpensive microphones which do little noise canceling. Good noise canceling microphones tend to score in the low twenties because their noise canceling technologies tend to muffle the sound slightly. Although the readings on the signal to noise ratio are lower, accuracy is always superior. Ergo, you cannot go by the signal to noise ratio readings excepting context of the type microphone and environments it is being tested (zero noise environment). When I say environment it is being tested, Dragon and other sellers of the Plantronics microphones do not do real-world testing. On the homepage of our http://www.eMicrophones.com is a link to, "Read How We Test Microphones"

Martin

I'm not familiar with this

I'm not familiar with this microphone or vendor. BTW, I wonder about the basis of your judgement that he is an expert? Prior purchases, familiarty with posts to on-line forums, or what?

SFAIK, Nuance is NOT bundling the Plantronics. The guy doesn't say that -- only that he tested it with DNS Medical 8. What else is there to test it with these days? For one thing, its too expensive for a vendor which is gaining a reputation of being pretty cheap.

Wherever you do buy, I would check out the return policy first in case the claims don't work out for you.

I'm not especially close to Martin -- we've had our differences over the years -- but I do buy microphones from him because I trust his judgement and am comfortable with his return and support policies. If he's not impressed by this microphone, I would be hesitant to buy it in the first place.

I also know and trust Graham Hendry and would feel comfortable buying something he recommends.

Both of these vendors are active and valued participants in this and other on-line SR communities, so they have to mind their P's and Q's or else the negative feedback would hurt their business. OTOH, I'm not sure the 21th Century guy contributes much here or other places I frequent.

One thing to be wary of is the fact that a mike can behave differently for different speakers. I don't like to be the first one to buy anything new because, well, because I'm that kind of guy. Often I find I get a positive first impression of something new based on the hype but later learn the product isn't all that well received by other users. I like to wait for that kind of feedback because I'm a little tight with my money.

This may be an OK mike at a good price, but I would suggest you'd be better served by a more skeptical approach to marketing claims not backed up by other sources. Shop around some more and check out some other venues, like knowbrainer.com/forum.

But if you do buy it, let us know how it works out -- both the pros and the cons.

Bruce

Some Comments on the

Some Comments on the Plantronics CS60-USB.

The UK version of the US CS50 is the CS60 (a different charging kit because of electrical supply differences). As a warning do check the designation - USB is the one required for speech recognition - there is also a telephony option with the same prefix but no USB.
Nuance are bundling the headset with DNS Preferred in the UK and some EU countries. See: http://www.scansoft.co.uk/news/20050830_dns8pref.a...
There was also a "limited edition" only available through VARs bundled with Professional.
The fact that Nuance bundle a product with NaturallySpeaking indicates that it is usable but doesn't necessarily mean that it is the best. There are obviously cost implications too. There have been many comments on this and other forums regarding the number of "dragon stars" issued to microphones and recorders on the Dragon compatibility pages. High ratings there don't necessarily correspond to user/VAR experience.

As Martin says an ASW value of 28 doesn't mean too much without considering other factors. I can increase the ASW score with almost any microphone (but perhaps not the CS60) by talking in a falsetto voice - but it doesn't do a lot for accuracy or the health of my voice (or my street cred.)! Also in a quiet environment any really experienced user should be able to get high 90's accuracy dictating a standard passage, even with a poor microphone, as long as the rest of the equipment is performing well.

I have tested the CS60 and found it to be reasonably accurate in a quiet environment, although not as accurate as the VXi BlueParrott TalkPro Bluetooth headset. The Plantronics headset is a fixed length boom of the type used in cell phones and consequently isn't adjustable to the users head size.

The bottom line is that some microphone resellers are very enthusiastic about the CS50/CS60 (KnowBrainer for one) others, including myself find it usable but not necessarily the best.
UK "street price" is around £169 exclusive of VAT as a comparison with your eBay price.

Graham Hendry
http://www.itspeaking.co.uk

Sigh . . . Half of the time

Sigh . . .

Half of the time I wish I were half as knowledgeable as I think I am the other half of the time.

Bruce

Thanks for your comments. I

Thanks for your comments. I am inclined to buy this mike now . It is the UK CS60-USB. Like you said, I was surprised 10 months ago when I got fairly good accuracy with a £5 microphone I bought for some other use. I will let you guys know how I got on if I decide to purchase this mike eventually
Builder

ignore the SNR from Dragon..

ignore the SNR from Dragon.. worthless, a nice parlor trick, that's about it.

I would love to have the wave file of this training session!
You achieved 99%, out_of_the_box, from a Plantronics CS-50 headset...?? Here, on planet Earth...?
[-G-]

And this was a brand new profile, or built/overlaid on top of previous training submitted from a normal/decent microphone..?

=================

Allow me to help a little w/ the Bluetooth discussion..?

It matters not what the microphone's Frequency Response is.. when it comes to digital wireless microphones. BT uses profiles, e.g "hands free", "headset", "modem", on and on. The BT.org SIG dictates what amount of datarate (read: voice bandwidth) can be transmitted under whichever profile.

There are *no* digital wireless microphones in production (and I say this from 9 months of research and about a zillion calls) that Tx past about 5.4khz voice bandwidth.

There are *no* BT devices that Tx past about 4khz voice bandwidth. the hands free profile demands a cap of 3.4khz and the headset profile, I think, is 4khz.

However:

The crucible is signal quality, for now. MS is going to/has raised the bandwidth bar here, but to **Dragon:... 8khz of voice bandwidth from a microphone with a washcloth over it vs. 3khz from a crisp, clear headset.. 'nuff said.

Until you get to the Sennheiser price range, the B 100 will outclass almost anything. Super mic element, super noise canceling, and the darn thing will pair to a lot of BT devices, so it's a multi-tasker, to boot.

PS: If one mentioned one was experiencing "dropouts" while using, say a B 100 to an application engineer at Broadcom or CSR they would most likely say something like.. "Well, you shouldn't be stuffing more than 50 or so active users into a small area, should you?"

** six months from now, I think, DNS CD's will be collecting dust on shelves everywhere.
. --> Get MS Speech Engine 6.1

http://www.microsoft.com/downloads/details.aspx
?FamilyId=5DAAE9C4-188C-4547-A9D6-1671132A39A1
&displaylang=en&EventType=getsdk
(admin: use http://tinyurl.com/dzy3n instead)

and train two sessions only, with a good (good defined as >6khz voice bandwidth microphone) microphone.. I'd wager a guess you'll begin wondering how to get your macros into your new speech recognition program.

.

Clarification..

Clarification.. oops..

.

Quote:

There are *no* BT devices that Tx past about 4khz voice bandwidth. the hands free profile demands a cap of 3.4khz and the headset profile, I think, is 4khz.

.

.
At least not yet,.. I'm watching for VoVox or one of the other headset makers who are using BT's EDR to raise the hardware bar soon..

jgradian wrote:hink, DNS

jgradian wrote:

hink, DNS CD's will be collecting dust on shelves everywhere.
. --> Get MS Speech Engine 6.1

http://www.microsoft.com/downloads/details.aspx?Fa...

and train two sessions only, with a good (good defined as >6khz voice bandwidth microphone) microphone.. I'd wager a guess you'll begin wondering how to get your macros into your new speech recognition program.

Umm, I'm wondering exactly how you're backing that guarantee, or if it is only a rhetorical device?

I've tried MS SR 6.1 and found it to be fairly accurate, but there was nothing there to motivate a desire to move my macros. In other words, if there's any cheese in that guarantee, I'd like to claim my share Smiling

Frankly, I don't think there's any room for any SR program to get significantly better than what we see in DNS 8.x and MS SR 6.1. Both programs use basically the same two-four decades old technology and aren't likely to show significant improvements until there's a new research breakthrough, which is an unlikely prospect according to the research-types I've canvassed.

We'll have to wait to see if Vista SR moves MS' offering up to the ease of use level that DNS now has. It does seem possible that Vista SR will gradually replace DNS, but only for the same reason that free beer replaces store-bought beer -- it ain't no better but its a darn sight cheaper.

Bruce

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.




view recent posts