I made several tests with different setups in order to achieve an acceptable speech recognition quality. It works well when I push a button to activate it but now I want it to be automatically activated when a user speaks. This is a big problem, especially when I only use the energy of the audio signal to guess when the user is speaking. That is why I thought about using a headset and not a distant microphone. In a headset the microphone is very close to the users mouth and it is easier to make correct guesses about when the user is speaking. Now my question is if bluetooth sets used with mobile phones also have such a property. They are not long enough and their microphone is not positioned exactly in front of the mouth. Is there a possibility that such devices can also capture some speech/noise from a distant user? Is there a significant difference in the signal energy coming from the user's speech and a 1 meter distant person's speech?
3 Answers
You might want to examine different microphone pickup patterns and determine which is best for your application:
Most headset microphones are designed using a cardioid or hypercardioid pattern for their noise-cancelling properties. They probably won't be as good for recording sound at a distance of 1 meter, but you can still get cardioid microphones that are useful at such a distance.
To understand better some of the design concepts of directional microphones, I recommend checking out EEVBlog video #605, which discusses it.
As to whether you should use a condenser, dynamic, ribbon, etc. microphone, that depends on the electrical characteristics of your device.
-
$\begingroup$ Actually my goal is to avoid as much as possible recording sound at a distance of 1m or even 50cm. The microphone should only record the speech of the user wearing it, few centimeters away from his mouth. $\endgroup$– MehdiJul 10, 2014 at 2:08
-
$\begingroup$ Definitely take a moment to watch the video I linked. The guest speaker explains how microphones can be designed for close proximity to the user, and cancel out other noises. $\endgroup$– JYeltonJul 10, 2014 at 6:30
-
$\begingroup$ Microphone arrays are the preferred way to recognize signals (such as speech) at a distance. These are available off the shelf. google.com/search?q=speech+recognition+microphone+array $\endgroup$– IanAug 5, 2014 at 1:32
Best audio needs mic be close, not 1 m away.
Consumer grade, 10 to 20 US dollars http://www.amazon.com/Panasonic-KX-TCA60-Hands-Free-Headband-Cordless/dp/B00007M1TZ/ref=sr_1_6?ie=UTF8&qid=1405009506&sr=8-6&keywords=headset
Professional Audio (stage, TV station, etc.) grade, 200 US dollars http://pro.sony.com/bbsc/ssr/cat-audio/cat-wiredmics/product-ECM322BMP/
Example for very best audio quality in real action http://www.ted.com/talks/amanda_palmer_the_art_of_asking
Hope it helps
-
$\begingroup$ I was actually looking for the same thing as the Panasonic but wireless for more convenience. I ended up buying the Logitech H800 which is maybe overqualified for only doing speech recognition. I enjoy listening to music while walking around my house though :D and I can recommend it. $\endgroup$– MehdiJul 11, 2014 at 2:41
Most laptops, cell phones, and conference phones actually use an array of microphones to properly locate the source of the person speaking, and filter out noise that comes from other locations. This is called "beamforming", used with a "microphone array".
See also: this answer.
