Incorporating a speech recognition quality microphone means that
the system described uses a microphone with a "flat" frequency reponse
across the frequency range of the captured speech. It also means more than
a good microphone element - it means superior microphone housing design
that avoids echo, and other unwanted acoustic distortions that occur
when even a good microphone is located in a bad housing design.

 A 'good microphone' will always perform good noise canceling; very good
accuracy from the microphone for speech recognition typically requires
unwanted background noise to be eliminated, so not to be 'heard' by
the recognizer and interpreted as speech sounds.

There are typically three types of noise canceling;
1. Using a microphone element with a super cardioid pickup-pattern;
2. Using mechanical noise canceling in the microphone houseing design;
3. Electronic or 'active' noise canceling in the circuitry & housing.

 Of the above three noise-canceling mechanisms, either (1) or (2) is usually
better. Speech recognition software typically tests signal to noise ratio when
doing a microphone "setup" or when respective setup wizards are run;
Whenever a new profile is started, for example, with Microsoft Speech Recognizers
(which the LOBE is optimized for) the wizard will test & calibrate the recording
device settings for the audio input device. "Active" noise canceling has tended
to confuse these wizards or cause false warnings, telling the user the
microphone is poor when in fact it may be quite good.

It's important to note that even when wireless systems incorporate
a "speech recognition quality microphone" element, meeting specifications;
If the captured voice bandwidth, from that microphone element isn't
successfully transmitted to the base unit & then sent off to the recognizer,
Even though a microphone is present that meets necessary criteria,
the system isn't necessarily a wireless system that will render good
voice delivery for accurate speech recognition.

The LOBE's headset microphones meets speech recognition standards in two areas;
it provides power-consumption friendly noise-canceling via both mechanical
and a basic cardiod pattern pickup-field, and has a rather flat frequency
response plot across the entire spectrum of speech bandwidth it captures.

The LOBE's headsets' wireless protocol is especially designed for, and is the only unit
presently able to, enable multiple users to transmit an 8khz voice bandwidth
signal in one close environment, as well as being the only commercially-available
unit designed to transmit a digital quality 16khz/8khz voice bandwidth signal.
Everything the microphone captures from the user's voice is sent.

Here's why:
Frequencies from about 200hz to about 5khz include commonplace speech data,
and higher frequency responses are necessary for true intelligibility.
Upper frequencies have the most significance for speech information
(often called speech data)about the phonemes (Wikipedia definition) you are saying,
for example spoken fricative consonants (Wikipedia definition) like [s], [z] and [3] for instance.

The frequency plot below is an example of the fricative consonant "S"
that is pronounced in ending the spoken word "Yes".
Clearly; most of the energy from this particular speaker's fricative "S" sound
doesn't really become significant until about 5.5kHz., which is well past the
'toll quality' ranges offered by other digital headsets/wireless microphones.
(Hertz frequency integers are bottom row)