bluetooth microphone suitable for voice recognition software?

Jonathan Epstein

unread,

Feb 26, 2003, 11:42:37 AM2/26/03

to

I am a user of voice recognition software, and am thinking about
buying a tablet PC if I can find a way to make voice recognition
software work with it via Bluetooth.

However, well-informed reports that I have read indicate that most if
not all Bluetooth headsets compress the signal when transmitting to
the receiver, thereby making them unsuitable for voice recognition
software. Reference:

http://support.lhsl.com/databases/dragon/webdisc.nsf/UNID/B391A6F7EBDFDF1585256C94004AA9C9?OpenDocument

Do you know of any bluetooth microphones which don't have this
problem?

TIA,

Jonathan

Jonathan Epstein

unread,

Mar 28, 2003, 4:22:16 PM3/28/03

to

Following up on my own message, I read today that 64kbps is the
highest data rate that one can use for a Bluetooth bidirectional
headset. Ref:
http://www.tomshardware.com/consumer/20030328/bluetooth-04.html

Now, voice recognition software such as Dragon NaturallySpeaking
requires 11.025kHz sampling, and expects (I believe) 16-bit samples.
So uncompressed you're talking about a requirement of 177kbps.

If this information is correct, this suggests that there will never be
a voice-recognition-quality Bluetooth bidirectional headset, although
this doesn't necessarily exclude a microphone-only device. Would any
bluetooth protocol expert out there please explain comment on this?

Thanks for any semi-technical pointers on this topic, including what
types of data compression might be supported in the context of a
Bluetooth device.

Parenthetically, I have been told by a vendor that the forthcoming
Bluespoon Digital headset will provide voice-recognition-quality, but
based upon the above analysis I tend to doubt it.
http://www.pilcon.com/

Jonathan

jaepst...@yahoo.com (Jonathan Epstein) wrote in message news:<27e4d2e8.03022...@posting.google.com>...

Martin Markoe

unread,

Mar 28, 2003, 5:11:24 PM3/28/03

to

Jonathan,

>>>>>>>>>>>>>>>>>>>>>>>>>>
So uncompressed you're talking about a requirement of 177kbps.

If this information is correct, this suggests that there will never be
a voice-recognition-quality Bluetooth bidirectional headset, although
this doesn't necessarily exclude a microphone-only device.
>>>>>>>>>>>>>>>>>>>>>>>>>>

We tested to Bluetooth microphones last summer. They were basically unusable
for speech recognition. On speaking to the manufacturers it was apparent
that the blue tooth microphones use compressed signals. They tend to clip
the beginning and end of what you say. The human voice cannot distinguish
this but it is deadly for speech recognition accuracy. The same holds true
for the 2.4 GHz microphones we have tested. Unless someone manufacturers one
of these microphones specifically for speech recognition is unlikely they
will work.

--
Martin Markoe, eMicrophones, Inc.
The best microphones for Speech Recognition
See us at: http://www.eMicrophones.com
See, "Key Steps to High Speech Recognition Accuracy" at:
http://www.emicrophones.com/docDetails.asp?DocumentID=38

Haru

unread,

Mar 29, 2003, 11:57:51 AM3/29/03

to

Hi, Jonathan.

--- CUT ---

> Thanks for any semi-technical pointers on this topic, including what
> types of data compression might be supported in the context of a
> Bluetooth device.

--- CUT ---

Although you may already know, the Bluetooth bidirectional
communication is using the codec type called "CVSD", Continuous
Variable Slope Delta modulation. Bluetooth headsets also use this
codec type on a bidirectional connection(64Kbps)

However, since I am not the specialist of a codec, I cannot give you
good advice. If you want to know more detail about "CVSD", please
search by using search engines like "Google".

I am sorry that I cannot give you help at all.

/Haru