Design Article: Cancel Noise In Your Mobile Phones And Headphones

I recently got a call from a colleague regarding a planned customer visit. He was hurrying through the Chicago airport amidst the noise of incessant boarding calls. He needed me to send out some documents before the end of the business day in Europe. Competing with all the background noise of the airport, I had to ask him to repeat himself and clarify exactly what he was inquiring about several times. It wasn’t until he entered the jet bridge that led to the airplane that I finally understood what the ordeal was about. 

Noisy environments are a part of our busy lives. Trains, planes, and automobiles are, for the most part, unavoidable. Your preferred restaurant is probably the favorite of many others and is no doubt filled with chatter and music throughout the dining experience. Poor voice quality provides an unpleasant experience for mobile phone users, and not only for the two people on each end of the line. Carriers must transmit the noise over the network, which in turn eats up more precious bandwidth in an already crowded spectrum. This can lead to more dropped calls, not to mention a less than favorable reputation for service providers.

Silent Solutions

Noise isn’t a new problem. Noise-cancelling headphones have been around for more than a decade to address what is known as ambient noise. Ambient noise cancellation deals with sounds that are constant, such as the hum in an airplane cabin. The technology works by creating a soundwave with the same amplitude as the incoming noise, but with an inverted phase (180°). When combined, the waves cancel each other out and the resulting soundwave may not even be audible to our ears (Fig. 1). 

Active noise cancellation produces an equal but phase-inverted version of the noise wave that produces almost complete cancellation when it’s added to the noise.

Ambient noise cancellation requires no or very little delay between noise capturing at the microphone and anti-noise generation at the speaker to match the anti-noise phase with the ambient noise phase at the user’s ear. This is a challenge for digital chip solutions, since the analog-to-digital (A/D) and digital-to-analog (D/A) conversions cause undesirable delay, and analog solutions are often used for ambient noise cancellation. However, some audio converters can be configured for very small A/D and D/A conversion delay and, therefore, can enable digital ambient noise cancellation solutions.

Continuous noise can be addressed with ambient noise cancellation, but suppressing rapidly changing audio signals, such as the noise we experience in airports and restaurants, is much more challenging. To address this issue, and for me to clearly understand my colleague running through O’Hare’s noisy airport, uplink noise cancellation is necessary. 

Several companies provide uplink noise suppression solutions on dedicated or multifunctional ICs, which can integrate the offering in hardware and/or software. The exact implementation details of uplink noise cancellation solutions vary by company and are considered the “secret sauce” for those working in this area. Nevertheless, some items must be considered no matter the secret sauce’s contents:

• Single-microphone noise cancellation: One microphone can be used to capture and effectively suppress the stationary noise without compromising voice quality. In a single-microphone environment, the same microphone will be used to capture voice and noise. The noise in the signal is suppressed by algorithms that examine the frequency spectrum and segment it into many channels. Each channel is analyzed by its amplitude characteristics.

• Dual-microphone noise cancellation: A two-microphone setup offers superior performance compared to single-microphone solutions, but it can be challenging for mechanics and acoustics. Dual microphones can be used to discern noise and speech signals. One typical approach involves analyzing both phase and amplitude difference as they arrive at each microphone. Another typical approach is to separate speech and noise by de-correlating those signals. Frequency-domain nonlinear processing may be necessary to suppress diffused noise. It requires noise-level estimation to determine the level of signal suppression in each frequency band.

• Microphone placement: In multi-microphone environments, special consideration should be given to microphone placement as a critical part of the system solution. For optimal performance, the speech level at the primary microphone is higher than the level at the reference microphone by at least >5 db in handset mode and >2 db in hands-free mode. This assumes the same level of sensitivity across both microphones. The reference microphone also should not be located near the speakerphone loudspeaker. Figure 2 shows some potential placements for the primary and reference microphones on a handset.

• Hardware/software solutions: While some solutions insist that noise suppression should be maintained in the analog domain, this limits the capabilities and variety of solutions that can be offered. A digital signal processing (DSP) architecture, which can process software algorithms, allows for precise tuning based on the acoustics of each system.

On The Market

Systems that integrate noise cancellation or suppression software into ICs that are already common on the board, like audio codecs, provide a great value in terms of space and power consumption. This type of hardware/software partitioning offloads the host processor and provides enough performance capabilities to run additional algorithms sequentially, like acoustic echo cancellation or other audio processing/filtering. To save power, the system can be programmed to turn off the noise suppression components in quiet environments. Many of these audio codecs even have added digital microphone support, which can provide better noise immunity. 

Fig.2: two-microphone noise cancellation works well when the microphones are placed in optimal positions.

Many competing solutions today embed a small DSP processor into the audio converter, typically a digital-to-analog converter (DAC) or a codec. These processors can vary in size, performance, and integration levels. Some integrate stereo or mono class D speaker amplifiers to provide a complete solution that can be tuned based on the mechanics and acoustics of the device. 

Many solutions provide a set of graphical software tools to create process flows and tune the system without the need for complex coding knowledge, as well as branded algorithms like SRS WOW HD that can be run on the device during audio playback. One such device is the TLV320AIC3256 from Texas Instruments.

In summary, finding solutions to the noisy environments of our daily lives doesn’t necessarily have to be a complex process or significantly lengthen the time-to-market of the end equipment. The key lies in finding a good partner/supplier with the right mix of silicon, audio/system expertise, and software.