Olympus ME-31 directional desktop microphone
We just finished testing the Olympus ME-31 Compact Gun Microphone. This is not an array microphone that will pick up up sound from all directions. It is like the shotgun microphone used by newscasters on site to pick up sound from just a specific narrow direction. It has standard 1/8" analog connectors. We tested with and recommend the Andrea USB-SA USB sound pod to assure the best audio input.
Frankly, I've never met an array or shotgun microphone that was really great for speech recognition. They just seem to pick up too much noise. Although I try to set my prejudices aside when testing a microphone, I have to say I really had no illusions this would be any good.
With no backgroound noise, I ran the Audio Setup Wizard in Dragon 10 and received 19 signal to noise ratio. There also seems to be a fairly low noise floor in a quiet environment. I then dictated the 313 word/commands standard selection, "The Rainbow Passage." When testing, I never look up as I wish to concentrate on enunciation. This is because I can routinely achieve 100% accuracy with the best microphones even with background noise. This is not a perfect method of testing, but over the years has worked well.
To my surprise, there were only five errors, a 98.4% accuracy rate with only ambient noise, no obvious noise. The microphone sits under my monitor, just under 2 feet away, pointing at a 45° up angle directly at my mouth. My keyboard is on the desk between my mouth and the microphone. With the up pointing angle the keyboard clicks when making corrections did not introduce extraneous words.
The next part of the test was in moderate noise. The radio was playing in the background and I had my product manager Matthew ring the phone. Although the microphone is directional, I suspect it was picking up reflections and reverberations as accuracy dropped to 94%.
In a quiet environment this is a convenient microphone. You need not put on or take off the microphone. In a quiet office it can sit up to 2 feet away and not be in your face like some desktop mounted microphones. CLICK HERE to see the Olympus ME-31. Click the, "click image to see larger version" link to see the microphone without the supplied windscreen.
As always, we recommend running the Audio Setup Wizard in Dragon or the Setup my microphon utility in Windows Speech Recognition. This does several things. It adjusts the audio levels, adjusts to background noise and it puts the person dictating into dictation mode where you remind yourself to enunciate clearly and speak in phrases.
Marty Markoe, eMicrophones, Inc.
Marty, It sounds like during
Marty,
It sounds like during testing, you had the mike lower than your mouth but still on the same y-axis plane as your mouth. How does it perform if it is, say about 45 degrees off the mouth vertical on the side of your monitor, so that you are not talking directly into it?
Stan
It sounds like
It sounds like during testing, you had the mike lower than your mouth but still on the same y-axis plane as your mouth. How does it perform if it is, say about 45 degrees off the mouth vertical on the side of your monitor, so that you are not talking directly into it?
Due to the nature of this type microphone, I had not thought of testing it off access before your suggestion. Before getting to this message, I dictated three e-mail, one of which was quite long. I had to make four corrections in the long e-mail. Only one keyboard click was picked up as a word.
The other microphone is about 3 feet away, not two feet as in the original testing, accuracy is still incredibly good. Even when the heating blower in the building turned on, accuracy was still high.
--
Martin Markoe, eMicrophones, Inc.
The best microphones for Speech Recognition
Read, "Key Steps to High Speech Recognition Accuracy.
Marty, You tested with the
Marty,
You tested with the Andrea full-duplex USB pod. Should I get the same results with the older plain Andrea USB pod?
Stan
snsok wrote: You tested
You tested with the Andrea full-duplex USB pod. Should I get the same results with the older plain Andrea USB pod?
There is no reason I can think of the Andrea half duplex USB sound card would not work. Every other microphone we have tested with the 1/2 duplex card works with the full duplex Andrea USB-SA sound card and vice versa.
--
Martin Markoe, eMicrophones, Inc.
The best microphones for Speech Recognition
Read, "Key Steps to High Speech Recognition Accuracy"
I am curious which of these
I am curious which of these two you would expect to be most accurate: the Olympus ME-31 or the Sennheiser VMX wireless...or would I not expect a whole lot of difference?
Stan
snsok wrote: I am curious
I am curious which of these two you would expect to be most accurate: the Olympus ME-31 or the Sennheiser VMX wireless...or would I not expect a whole lot of difference?
In a quiet environment they would be about the same.
--
Martin Markoe, eMicrophones, Inc.
Our Comparison Results
We have compared the Senneheiser VMX Office to the Olympus ME 31 and also found them to be about equal in accuracy but keep in mind that you're comparing a wireless dual microphone element headset microphone to a desktop directional microphone, which is an apples and oranges comparison. Additionally, note that if you're going to back off more than 12 inches, we do not recommend dictating into the ME-31 off axis because it is a directional microphone which is specifically designed to ignore off access sound. The main reason for dictating off axis is to avoid windage on the microphone element but if you're dictating from 6 or more inches away, windage will not be a mitigating factor. The rules for utilizing a directional microphone are different from a standard headset and you won’t need to use the included wind screen unless you're dictating outside or under a brisk fan. One of the first things we noticed, about the ME-31 microphone, is that you can dictate fairly accurate from up to a meter or more with the proverbial sweet spot being between 7 to 30 inches.
We will be posting our entire Olympus ME 31 Review on the Microphone Forum and have ME-31 available now (with our new ME 31 training guide). We apologize for not having done so sooner, but we are firm believers in doing extensive testing of a new product rather than rushing it to market.
Lunis Orcutt - Developer of KnowBrainer &
Host of the http://www.TheMicrophoneStore.com
A Nuance Gold Certified/Endorsed Partner
ALWAYS Ask If Your Speech Recognition/Microphone Vendor Is Nuance Certified
KnowBrainer wrote:we are
we are firm believers in doing extensive testing of a new product rather than rushing it to market
Hmmm??? Since the inception of our web site specializing in speech recognition microphones 12 years ago we have posted a link on our home page that says, "Read How eMicrophones Tests Microphones." In addition, many of our microphone pages do not require you to read a manual of users guide as they have 1 or 2 minute movies to explain everything in an easy to follow visual format.
As for many years Knowbrainer added microphones to their web site based on the testing and innovation of eMicrophones we are happy to hear you are now actually testing microphones in your home office environment. eMicrophones office and shipping facility is in a commercial office builing. eMicrophones tests microphones in quiet, moderate noise and high noise environments that are akin to what real life users (home office, private office, legal office and noisy medical facilities) encounter on a daily basis.
By the way, our testing shows using the supplied windscreen on the Olympus ME31 Compact Gun Microphone is an advantage in most dictating environments.
--
Martin Markoe, eMicrophones, Inc.
The best microphones for Speech Recognition
Read, "Key Steps to High Speech Recognition Accuracy"
Martin Markoe wrote: We
We just finished testing the Olympus ME-31 Compact Gun Microphone. This is not an array microphone that will pick up up sound from all directions.
How about the clicking of a keyboard? Does the microphone pick it up? what would you say is the beamwidth (e.g. spotlight size) of the microphone on target if it is about 18 inches away. In other words, does the spotlight cover your entire face? Can you tell?
reckoner wrote: How about
How about the clicking of a keyboard? Does the microphone pick it up? what would you say is the beamwidth (e.g. spotlight size) of the microphone on target if it is about 18 inches away. In other words, does the spotlight cover your entire face? Can you tell?
In our testing, the microphone was sitting between the keyboard and the monitor pointing up at our mouth. The keyboard was underneath and in front of the microphone. It did not pick up keyport clicks as extraneous words. I would have to guess that the spotlight size starts out narrow and spreads out like a funnel the further away it gets. Unless the microphone was pointed very low towards the keyboard, I doubt it will pick up keyboard clicks.
Just to reiterate, this microphone is very useful in a quiet environment but not even with moderate noise like music playing in the same room.
--
Martin Markoe, eMicrophones, Inc.
The best microphones for Speech Recognition
Read, "Key Steps to High Speech Recognition Accuracy."
Here We Go Again
Marty, we don’t know what set you off this time but if we must do this again…
As for many years KnowBrainer added microphones to their web site based on the testing and innovation of eMicrophones we are happy to hear you are now actually testing microphones in your home office environment. eMicrophones office and shipping facility is in a commercial office builing. eMicrophones tests microphones in quiet, moderate noise and high noise environments that are akin to what real life users (home office, private office, legal office and noisy medical facilities) encounter on a daily basis.
Giving you the benefit of the doubt, we are probably about equally split when it comes to one of us introducing a microphone before the other. We have been professionally involved with marketing and developing speech recognition software and peripherals (microphones, digital recorders and command software) since 1993 and have maintained a vendor web presence since 1996. It sounds like you really don't know how we test microphones so we will bring you up to snuff. Depending on what microphone we are testing, we have been known to use up to 20 different testers but we typically only use between 5 and 7. We use a combination of in-house and independent testers. We do most of our work out of our home office but we see no advantage in utilizing a commercial office building. We consider a commercial warehouse to be a waste of space, a waste of time (driving to it) and non-environmentally friendly. We will be moving to a much larger facility this spring, but will still be working from our own property, (34 acres of woods and fields, which besides our new home will include approximately 4000 sq feet of work space).
The following is how we test microphones:
Phase 1 testing is conducted with a well-trained user profile to see if continued testing is warranted. Some microphones do so poorly that they never go beyond phase 1 testing. Our most recent failure was the Sennheiser PC 21 headset which we found to be slightly uncomfortable and turned in an accuracy score of 96%. Except for highly specialized microphones, our accuracy cut-off point is 97%. We also have to take SRP into account. The Sennheiser PC 21 made it to phase 2 testing because it only retailed for $30. Although the PC 21 didn't quite make our recommendation list, for $30, it would still make a good starter microphone and we may revisit it in the future.
Phase 2 testing is comprised of a new pristine user profile that is never saved. We test for accuracy, noise cancellation and comfort but our tests are always conducted by reading pre-fab text which is equivalent to “testing in a vacuum”. We never read familiar text like the Rainbow Passage. Whenever you read text into NaturallySpeaking, it's going to be notably more accurate because free-form dictation tends to be choppy. After reading the same text, 3 times, we replace the text because each time you read the same text into NaturallySpeaking it will become more accurate (without correcting) because you start to memorize the text and your voice gets even smoother. Each training text is 300 words and we perform 3 separate tests for a total of 9 readings. Although this is a rough estimate, we recommend deleting 2 full percentage points from our accuracy scores for a reality check (typical free-form dictation). Although we average the scores between all testers, a microphone that does very well for one person may not do so well for another. Before releasing our results, we additionally go through phase 3 testing which is not a measured score.
Phase 3 testing is designed to compensate for the placebo effect as we sometimes fall victim to shiny objects. Even though we have finished our review by the end of phase 2, we simply can't trust our results until they undergo real-world testing which means using the microphone for all our dictations for 2 weeks. It's amazing how may times we run into something that wasn't obvious in the 1st 2 tests or even the 1st few days. For example, we found the Nady Encore II had a tendency to leak audio into NaturallySpeaking which resulted in NaturallySpeaking refusing to acknowledge our dictation for about 10 seconds after turning the microphone back on. Sometimes it's the little things that you don't notice until you seriously commit to using a microphone. Sometimes a system that seems perfectly comfortable during early testing turns out to be uncomfortable in the long run.
By the way, our testing shows using the supplied windscreen on the Olympus ME31 Compact Gun Microphone is an advantage in most dictating environments.
We have to disagree with you on this one. Our tests show absolutely no advantage with the wind screen and we will be recommending against its use in most situations.
I hope this clarifies things for you and puts an end any further snarkyness on this thread.
Lunis Orcutt - Developer of KnowBrainer &
Host of the http://www.TheMicrophoneStore.com
A Nuance Gold Certified/Endorsed Partner
ALWAYS Ask If Your Speech Recognition/Microphone Vendor Is Nuance Certified
KnowBrainer wrote: Marty,
Marty, we don’t know what set you off this time but if we must do this again… I hope this clarifies things for you and puts an end any further snarkyness on this thread.
What set me off was your "snarkyness" when you made the statement, "we are firm believers in doing extensive testing of a new product rather than rushing it to market." As I was the person who introduced the ME-31, the implication was taken personally and with merit.
Good luck with your move from the house backing up on the strip mall. After all those years it must be a relief. All the pollutants (noise, dirt and toxic carbons) is known to affect physical and mental health.
--
Martin Markoe, eMicrophones, Inc.
Nothing to Do with You Marty
What set me off was your "snarkyness" when you made the statement, "we are firm believers in doing extensive testing of a new product rather than rushing it to market." As I was the person who introduced the ME-31, the implication was taken personally and with merit.
We don’t know why you thought this had anything to do with you Marty; not everything is about you. If you re-read our posting you will realize that we were just apologizing for being slow and sometimes overly thorough. We neither said nor implied anything about your work habits, only our own.
Lunis Orcutt - Developer of KnowBrainer &
Host of the http://www.TheMicrophoneStore.com
A Nuance Gold Certified/Endorsed Partner
ALWAYS Ask If Your Speech Recognition/Microphone Vendor Is Nuance Certified
KnowBrainer wrote: Marty,
Marty, we don’t know what set you off this time but if we must do this again… I hope this clarifies things for you and puts an end any further snarkyness on this thread.
What set me off was your "snarkyness" when you made the statement, "we are firm believers in doing extensive testing of a new product rather than rushing it to market." As I was the person who introduced the ME-31, the implication was taken personally and with merit.
Good luck with your move from the house backing up on the strip mall. After all those years it must be a relief. All the pollutants (noise, dirt and toxic carbons) is known to affect physical and mental health.
--
Martin Markoe, eMicrophones, Inc.
At a typical 45° angle,
At a typical 45° angle, key clicks will pretty much be ignored unless you're one of those folks (not mentioning the Big Haired Lady by name) who pounds on the keyboard.
Martin is partially correct about the behavior of directional microphones. From 18 inches away you can actually get away with turning your head from side to side because you are close enough to the microphone to be picked up but the further away you get, the more you need to face directly down the barrel of the microphone. From 3 or 4 feet away, this becomes critical.
Lunis Orcutt - Developer of KnowBrainer &
Host of the http://www.TheMicrophoneStore.com
A Nuance Gold Certified/Endorsed Partner
ALWAYS Ask If Your Speech Recognition/Microphone Vendor Is Nuance Certified
Marty: how does this compare
Marty:
how does this compare to the Buddy Gooseneck USB Microphones you sell? Is the main difference the amount of stand-off. It seems to me that these Gooseneck mics would be more sensitive to head position and easier to bump into because of that.
comments?
reckoner wrote: how does
how does this compare to the Buddy Gooseneck USB Microphones you sell? Is the main difference the amount of stand-off. It seems to me that these Gooseneck mics would be more sensitive to head position and easier to bump into because of that.
The Buddy Gooseneck, Desktop and Flamingo are slightly more noise canceling and thus slightly (about 1%) more accurate.
Yes, stand off is the main difference.
Contact us through the Contact Us page of our web site and we'll arrange for you to try one. Just mention you are reckoner from Speech Computing.
--
Martin Markoe, eMicrophones, Inc.
Although we agree with
Although we agree with Martin that the Buddy Gooseneck, Buddy Desktop and Buddy Flamingo are more noise canceling then the Olympus ME 31, we disagree about the accuracy. In our tests, the ME 31 came out on top for accuracy.
Lunis Orcutt - Developer of KnowBrainer &
Host of the http://www.TheMicrophoneStore.com
A Nuance Gold Certified/Endorsed Partner
ALWAYS Ask If Your Speech Recognition/Microphone Vendor Is Nuance Certified