We’ve spent a significant amount of time on our blog in recent weeks focused on active noise cancellation (ANC), from how ANC technology works and the three basic types of ANC to multimicrophone noise-cancellation technology preventing undesirable sounds from getting into a voice signal. And that topic trend continues with this post, which dives into different active noise cancellation intellectual property (IP) options and how to determine which option is best, as well as tips, tricks and best practices for implementing ANC into products.
In recent years, ANC technology has been integrated into an increasing number of audio platforms at lower price points. From headphones and earbuds to smartphones and automotive applications, numerous use cases call for active noise cancellation, and the constantly growing number of different ANC IP options available on the market makes these integrations possible.
Chips, SoCs, Processors and Software: Different ANC IP Options
When it comes to active noise cancellation, or active noise reduction (ANR), some companies include active noise cancellation in their system on a chip (SOC) that they’re integrating into a device, while others develop custom algorithms that they run on separate digital signal processors (DSPs) built into their device. Plus, some companies even take noise-cancellation algorithms and run them on a general-purpose processor inside the device to cancel out unwanted ambient noise.
Of course, there are tradeoffs with each of these different design options. For example, if you’re using a built-in, off-the-shelf ANC software solution, it’s important to understand how it works. When using licensed software IP, you’re generally limited by where the algorithms will run and how the solution is packaged. Typically, companies developing off-the-shelf ANC software solutions are going to give you the code in binary form, and it will be built for specific processor targets — there might be an ARM version and versions for specific DSPs — and that’s where you get to run it. However, if you’re purchasing a significant number of ANC solutions, the company developing that IP will likely be happy to generate a binary for you for whatever DSP you choose to use.
But if you develop your own ANC technology solution or purchase one from another company, knowing where it runs is crucial. By generating your own code, you can run active noise cancellation on any general-purpose processor or DSP you see fit, with no limitations. You’re either creating your own models and running them anywhere you want or whoever is providing the models is giving you binaries that run on specific targets.
This delicate balance impacts how much memory is needed, how much power will be used and much more, which makes determining the design options that will deliver the most value for your product a matter of utmost importance.
The Best Options for Designing Products with Active Noise Cancellation
So, how can you determine what the best ANC technology solution is for your product? Well, the best options are only going to be “better” than others in the context of the application.
For example, if you’re developing truly wireless ANC earbuds, regardless of any other factors, the ANC IP option you select must fit on the hardware that will fit in the form factor of the earbud.
On the flip side, what makes the lower-quality options bad? Simply put, when a user is trying to listen to an audio signal and the music or voice coming through doesn’t sound good, that’s suboptimal. There’s no way to do noise cancellation without touching the music/audio signal going to your ears. So, if you utilize the wrong ANC chip, you run the risk of introducing artifacts into the audio signal that would range from annoying to unpleasant.
Why Noise-Cancellation Software from an Audio Company is Best
While audio companies don’t develop better proprietary noise-cancelling algorithms for their IP options per se, engineers at these organizations know and care about sound, and they know what music is supposed to sound like. As a result, they are often better able to hear the effect of the noise cancellation, ensuring the algorithm is doing the right thing and that the audio sounds good.
At the end of the day, when you go for an IP option from a higher-end audio organization, you’re going to get noise-cancellation technology that actually works, the music will still sound like music and your product is going to get audio quality from people who have a “sound.” Plus, audio companies tend to have the most sophisticated ANC technology on the market, often with more processing horsepower in a smaller power envelope than alternative options.
Tips, Tricks and Best Practices for Implementing ANC IP Options
The single most important factor is that you cannot ignore the mechanical design of your device. From the ear seals to the physical size of your product, you have to do the acoustical modeling on the device — otherwise, there’s no point.
If you don’t get the modeling right so that the DSP filters have the right parameters, at best, the ANC algorithm is doing nothing. Plus, if your filter coefficients are incorrect, you’re going to have something messing with the sound of the music.
Let’s say your earbuds have a recessed microphone that is not dealt with properly. That microphone’s sound will have either a notch or a ring, so the frequency spectrum will either have a dive or a hump in it. The noise-cancelling algorithm in the IP option you select is then trying to deal with the sound energy in that frequency band that isn’t there or will either allow some noise to leak through inappropriately or will change the tone of the music in that band in an inappropriate manner, sacrificing audio quality.
As a consequence of not doing the appropriate acoustical modeling for the recess on the microphone, when an end user is listening to music, the singer’s voice might change from a low to a high note, and at some point, the volume of the singer’s voice would either increase or decrease. The audio signal going into a user’s ear will hit certain frequencies where it’s wrong, and they’ll be able to notice a difference. There’s going to be standing waves of audio on the surface that the mic is on, and if you’re not modeling that appropriately, you’re going to have the same issue, except it will be across a wider frequency band. So, instead of hearing ringing or some other weirdness, the audio signal will sound muddy. Even worse, without appropriate acoustical modeling, you might not even get the noise-cancellation aspect correct.
Factors to consider include acoustical modeling, on-head modeling and other considerations, such as the actual characteristics of the specific microphone you’re using, the microphone’s placement on the device, where standing waves are occurring on the device and whether the mic is flush with the surface of the device or how recessed it is. You rarely want the microphone flush on the device because that way it’s exposed to damage. But when recessed a bit, the recess creates in itself a filter that has to be modeled in and dealt with.
For the most optimal noise reduction, it’s important that the ANC technology solution you choose is cancelling the specific frequencies that are creating the noise. If you’re not dealing with standing waves well, your ANC solution detects the frequencies wrong, and the audio not only sounds crappy but the ANC algorithm won’t protect your ears from undesired noises.
If you don’t tune your algorithms, the noise-cancelling algorithm might help save a user’s hearing, yes, but it won’t contribute to them hearing what they’re supposed to through earmuffs or earbuds. And if you’re using ANC technology for an application in which someone will be listening to music, and you don’t do the correct modeling, then when the user turns on the noise cancellation, it sounds worse because the noise-cancelling system will alter the music signal to deal with noise outside of the passive ear protection.
Our Active Noise Cancellation Expertise
At Cardinal Peak, we get ANC. We understand that if you want to design and build products with active noise cancellation, you have to tune the algorithms to your specific device. Our engineers excel at modeling and implementing algorithms in constrained environments, and our deep expertise with ANC ensures we can help you achieve the highest level of performance. Get in touch with us today to discover how we can help you determine which ANC IP option is best and implement the right solution for your unique product.