Fundamentals of Ambient Noise Cancellation for Headphones and Handsets
Silence is a precious commodity. Unfortunately we can not control it. All of us have experienced trying to make a mobile phone call from a noisy street, crowded restaurant or train station where the background noise can make it impossible to hear the incoming call. There are numerous noise cancelling solutions that filter and improve the voice signal from the microphone and create a high quality transmission from you to the other person on the call. But receiving a call in a noisy place can make the voice of the person calling you unintelligible.
Ambient Noise Cancellation (ANC) Basics
The basis of electronic ambient noise cancellation is destructive wave interference between the unwanted noise signal and a “cancellation” signal that is created to be equal in magnitude to the noise signal and opposite in phase to it. This is shown on the left of Figure 1, which depicts a sine wave noise signal and its cancellation counterpart and also the remaining residual noise that results from this interaction. The cancellation is perfect and the residual noise signal is zero. This seems like a simple approach, however it is interesting to consider what happens when the amplitude and phase characteristics of the unwanted noise and cancellation signals are not perfectly matched.
As an example, assume that the phases were perfectly matched but the amplitude of the cancellation signal was 20 percent smaller than the amplitude of the noise signal, then the cancellation would be incomplete and a significant residue signal would remain. This is depicted in the center of Figure 1. Similarly, if the amplitudes of the signal and the noise were perfectly matched but there was a 10 percent phase mismatch as shown in the right of Figure 1, then the cancellation would, again, be imperfect and the resulting residual noise signal would be noticeable.
The noise cancellation mechanism fails unless both the amplitude and the phase properties of the signals are closely matched, simultaneously, and this condition must prevail throughout as wide a range of frequencies as possible. Very tight signal matching is needed for even a modest amount of noise cancellation. If 65 percent cancellation (9 dB) is to be achieved (that is, the residual noise signal would be 35 percent of the original amplitude), then, assuming perfect phase matching, the amplitude of the cancellation signal must be matched to that of the noise signal within ±3 dB. Similarly, even if the amplitudes were perfectly matched, the relative phase of the signals must lie within ±20° for 9 dB cancellation which, at 2 kHz, corresponds to a time period of only 28 µs, and represents an acoustic path length of only 10 mm.
Feedback and Feed-Forward Systems
There are two different approaches for providing ambient noise cancellation for headphones and handsets: feedback and feed-forward. The feedback method uses a miniature microphone placed directly in front of the speaker pointing at the ear and is coupled back to the speaker using a negative feedback loop. The microphone senses the music (or voice) and the noise. The electronics separate the signals and generate an anti-noise signal that is added to the music such that it effectively cancels out the noise so that only the music is heard by the user. Although this principle is simple, there are some practical problems relating to the intrinsic phase response of the loudspeaker and the propagation delay between the speaker and microphone, which both introduce a phase lag at higher frequencies. Consequently, high-frequency filtering must be introduced into the feedback loop, which tends to restrict the upper frequency response to around 1 kHz or below. Feedback systems can be effective at canceling low-frequency noise but the acoustic seal needed can make the headphones bulky. Feedback-based ANC can be useful in applications such as “over the ear” or intra-concha headphones where the headphone can help seal the ear from outside noise.
For applications such as handsets or lightweight, small headphones, the feed-forward noise cancellation method may produce better results. In feed-forward systems, a microphone is placed at the exterior of the headphone shell in order to detect the incoming, ambient noise signal, which is then inverted and added to the headphone drive signal, thus creating the cancellation signal. The feed-forward system does not require such a well sealed cavity around the ear.
The feed-forward method is very simple to implement in a basic form, but current systems are far from perfect and generally limited in their effectiveness for several reasons. One fundamental difficulty is caused by the differing acoustic pathways that the noise and the cancellation signals travel which results in differing frequency-dependent amplitude and phase variations on the two signals. This requires careful signal processing to re-balance and match the two signals as closely as possible. Unfortunately, the use of conventional digital methods is ruled out because even a single A/D sampling period could introduce a 30 µs delay, and this alone would significantly disturb the phase alignment of the signals.
There is a further major difficulty in trying to time-align the signals: the acoustic path length difference between the ambient-to-ear and ambient-to-microphone varies according to the direction of the noise source. For example, if the microphone inlet lies at the center of the outer shell, then a wave-front arriving from a frontal (or rearward) source would arrive at the microphone and the entrance to the ear canal at approximately the same time. However, the cancellation signal then has an additional effective path length of three or four centimeters to travel to the ear due to the response time of the speaker and its distance from the ear. This corresponds to time periods of 87 µs to 117 µs, causing large phase differences that would prevent effective noise cancellation.
Time-aligned Noise Cancellation
A new method has been developed of overcoming these timing problems to provide a time-aligned feed-forward noise-cancellation system. The invention entails the use of a distributed array of microphones around the headphone or handset. Each microphone represents a notional noise-leakage entry point around the shell, and hence the sum of their signals represents the summed sound pressure level around the shell which is the driving force behind the incoming leakage. Because the headphone shell acts as a baffle, the acoustic leakage pathway from ambient-to-eardrum is forced to traverse one-half of the diameter of the earphone assembly before reaching the auditory canal axis. Accordingly, by placing microphones at the rim of the headphone, the ambient noise signal can be acquired and driven to the loudspeaker in advance of its arrival at the eardrum, thus compensating for the intrinsic response time of the loudspeaker. Furthermore, because of the distributed geometry of the microphone array, this applies to wave-fronts arriving from all directions and hence the system is largely direction independent.
This time-alignment noise cancellation technology has also been applied to mobile phone handsets in order to generate a noise-cancellation signal via the handset’s built-in speaker, thus creating a zone of relative silence at the listener’s ear, and improving the intelligibility of conversation.
Figure 2 shows a simplified block diagram of a handset with ambient noise cancellation based upon the Wolfson WM2000. The WM2000 Speaker Driver with Ambient Noise Cancellation is placed just before the speaker output. Two additional microphones are used for the noise cancellation. This implementation features wide cancellation bandwidth of 300 Hz to 2.5 kHz and provides 20 dB of ambient noise cancellation, assuming well designed handset acoustics.