Using a cellular phone in your car can involve a trade-off between safety and intelligibility. Holding the phone to your ear requires you to take one hand from the wheel causing serious safety implications. Hands-free car phones are a safer alternative; however, they typically offer poor conversation quality.
But significant improvements in the quality and intelligibility of hands-free car phones through the application of digital signal processing (DSP) techniques are now possible.
Hands-free car kits for in-car mobile phone use are offered across a spectrum of speech-processing quality, performance, and design. In its smallest form, a car kit can be manufactured as a "zero install" unit. These transportable units are typically a single device with an integrated loudspeaker and microphone and are conveniently powered by insertion into the cigarette-lighter adapter of a vehicle.
"Permanently installed" car kits which usually consist of a separate loudspeaker, microphone, junction box, and antenna, are designed to be permanently mounted into the vehicle, often by a qualified technician. The audio quality of these permanently installed systems is often better than achievable from a typical zero install unit, in part due to the use of higher quality components and an antenna. However, both varieties of kits are susceptible to noise and echo.
While the interior of a car can solve some speech intelligibility problems by eliminating extraneous traffic noise, it also creates problems by containing noise inside the vehicle. A closed passenger cabin of an automobile acts somewhat like an echo chamber and creates serious drawbacks in conversation quality.
These drawbacks to hands-free automotive communications are largely a result of the remoteness of the microphone from the driver's mouth and the distance of the loudspeaker from the driver's ears. This often results in the free-flow of conversation being hampered by unwanted background noise picked up by the microphone and transmitted to the person on the other end of the phone connection (far end).
Many hands-free kits utilize muting to prevent the far end speaker from hearing his own voice (acoustic echo) as it comes out of the loudspeaker in the car and is picked up by the microphone. But this solution permits only one person to speak at a time and can make the conversation seem unnatural.
The use of sophisticated DSP techniques can, however, greatly improve the naturalness and quality of the conversation by effectively enabling both parties to talk at the same time while preventing acoustic echo and also by reducing background noise. Such DSP techniques have been around for some years, yet it is only as the cost of DSP chips has decreased that consumers have been able to benefit from these advances in automotive electronics technology.
Noise reduction techniques
Even normal intelligible conversation between two people situated in a noisy environment can be very tiring. But the issues are compounded when communicating via a mobile phone with a remote microphone. As show below, the remote microphone of a hands-free car kit picks up not only the voice of the person in the car, but also the noise from the engine, wind, tires, fan heater, and other external noises. This noise contaminates the signal fed to the phone which is expecting a voice signal.
Inside the mobile phone, the signal is digitally coded before being transmitted. At the receive end, the signal is decoded back into speech. If the signal to the phone contains noise and speech, the received and decoded signal may be distorted since the unwanted noise could be interpreted as 'speech' by the coding algorithm.
A simple method for reducing the noise is to use a high-pass analog filter that cuts out the low-frequency noise picked up by the microphone in the car. Such a technique however, does not allow noise with the same frequency as the speech to be removed since the filter has no way to distinguish noise from speech. Using more sophisticated signal processing techniques can make the distinction possible.
An early example of such processing involved monitoring the microphone signal strength and subtracting the minimum signal, which was assumed to consist of noise without speech, from the rest of the signal.