Noise Cancellation Advances with AI
Hundreds of phones are ringing and agents are talking to customers and banging away on keyboards at once; there’s constant background noise from office equipment; and large heating/ventilation/air-conditioning units are running nonstop. Amid these and many other distractions, contact centers can be very noisy environments. And then when the COVID-19 pandemic sent agents to remote work environments, noise became even more of an issue, with children, pets, road traffic, and other typical household disruptions further degrading call quality.
Excessive noise levels in contact centers can impact operations, affecting communication and productivity. IRIS Audio Technologies notes that nearly 60 percent of customers have hung up on customer service calls due to background noise.
The company says that excessive noise has the following negative impacts:
- An erosion of customer trust because it can make the contact center seem unprofessional, especially if calls involve sensitive details like insurance claims, medical information, etc., that require privacy and confidentiality.
- Negative customer experiences, as customers can feel stressed by having to talk over the background noise and struggle to hear the agent. This can cause anxiety, especially for customers who are already frustrated when they call.
- Poor employee experiences, as agents feel ineffective. Nearly 9 in 10 contact center employees say noise hurts their performance levels, with 85 percent saying they waste time repeating information.
- Higher employee turnover. A recent survey of call center agents found that background noise is directly linked to higher turnover rates—30 percent to 45 percent annually.
Beyond the employee and customer concerns, in some locations there are regulatory concerns about noise levels. Several countries around the world have enacted legislation that sets the maximum average and peak noise levels that workers may undergo and the maximum amount of time that workers can experience these sounds.
Still, noise is an inevitable reality, and previous attempts to solve for it have had very limited success.
“Contact centers have never been quiet places,” says Christian Hed, chief marketing officer of Dstny, a provider of cloud-based business communication solutions. “Background chatter, keyboard clatter, the hum of fluorescent lights—it’s a symphony of distraction. For years, the go-to solution was basic noise suppression with static filters that muffled everything indiscriminately, including the agent’s actual voice. Useless.”
Two years ago, noise suppression meant low-pass filters chopping out everything above a certain frequency, Hed adds.
Some more advanced technologies offered more advanced capabilities, including active or passive noise cancellation, that was better but still fell short.
Active noise cancellation (ANC) works by using microphones to pick up external noise and then generating sound waves to cancel it out, essentially blocking it. Passive noise cancellation, on the other hand, physically blocks out noise through the design and materials used in contact center headphone construction; this includes thicker padding or tighter-fitting ear cups to reduce the amount of external noise that can reach the ears.
But those solutions were still static in that they didn’t evolve with the user’s environment.
Advanced Noise Cancellation Advances
So increasingly contact centers and others in need of noise reduction, such as salesmen on the road, for example, are increasingly moving on from the older noise suppression technologies and instead relying on advanced noise cancellation (ANC) technology. For headphones and earbuds used by those outside of contact centers, ANC technology has become the best way to eliminate unwanted noise. This technology is no longer reserved for the most expensive headphones, notes a SoundGuys blog, adding that there is a decent selection of earbuds and even true wireless products sporting some form of ANC technology. However, not all ANC implementations are equal.
There are numerous ways to implement ANC technology, each with its own implications for the quality and type of noise that the headset can address, the blog adds.
ANC uses a principle known as phase cancellation. Sound waves that are 180 degrees out of phase, or the inverse of one another, cancel out when summed together. The idea with noise cancellation is to record the background noise, invert the noise signal to create anti-noise, and then add it to your output signal. The anti-noise signal cancels out the actual background noise at the point it reaches one’s ear.
The biggest issue with ANC is sampling ambient sounds accurately enough to provide the maximum degree of attenuation, according to SoundGuys. “Microphones must capture the noise, and the phase of the cancellation waveform leaving the headphone drivers needs to perfectly line up with the phase of the noise when it reaches your ear. These systems need to be finely tuned, but even then, you won’t ever see 100 percent cancellation. Instead, between 20 decibels and 40 decibels of noise reduction is quite common, which cuts the background noise level you hear to between one-quarter to one-sixteenth its original level.”
The Move to AI-Based Noise Cancellation
The basis of ANC technology is 90 years old, according to SoundGuys. Today’s contact centers that have implemented the best noise cancellation rely on much newer technology that uses artificial intelligence to filter out the unwanted sound without canceling the sound of a soft-spoken agent, inflections in voice, etc.
“AI flipped the script,” Hed says. “Instead of brute-force suppression, it uses deep learning models trained on millions of voice interactions, allowing it to distinguish between actual speech and background chaos.”
And though the AI-infused technology is still very new, Hed has already seen benefits in several use cases:
- Mansjouren Stockholm, a crisis helpline, had volunteers struggling to hear distressed callers over background noise. After implementing AI-powered voice isolation, they tripled the number of calls they could handle, meaning more people got the help they needed.
- Svanströms Electrical & Plumbing, also based in Stockholm, saw a 60 percent drop in call disruptions after integrating AI-driven interactive voice response and noise suppression, leading to double the customer orders.
“Bad audio is bad business,” Hed says. “Companies that fail to adapt to AI-powered voice clarity are losing customers, revenue, and trust. If every call matters, then so does every word. AI-assisted systems make absolutely sure that only the right ones get heard.”
“Companies that have implemented this software have seen improvements in agent performance and reduced stress levels, alongside increased customer satisfaction,” agrees Smitha Baliga, CEO and chief financial officer of TeleDirect Communications, a contact center outsourcing services provider. “Key performance indicators such as average handle time have led to better first-call resolution rates. Generally speaking, clearer call recordings lead to more accurate quality control, increased productivity, and reduced listening fatigue that comes with too much distracting background noise.”
“Costumer service hubs have begun using AI programs to make user experience seamless,” adds Jodi Miller, senior vice president of sales at MAP Communications, a provider of call center and answering services. “For us, AI solutions have added clarity to our communication. Natural language processing has assisted us to understand and respond to customer questions without extra hassle.”
Rather than cutting off all sounds outside of a specific frequency range, as noise suppression does, Hed explains that AI-based noise cancellation does the following:
- It learns what’s actually noise. The AI is trained on real-world conversations so it knows the difference between an agent’s voice and the guy in the next cubicle microwaving his lunch.
- It filters out distractions in real time. Unlike traditional noise cancellation, AI actually adapts to different environments. It can handle open offices, remote home setups, and even call centers with more than 100 agents talking at the same time.
- It boosts voice clarity instead of just suppressing noise. AI sharpens the audio so that speakers sound like they’re in a quiet, controlled environment, even if they’re not.
The AI component means that noise cancellation precision continues to evolve, Baliga says. “AI-driven tech can now handle a wider range of noise profiles, with some being able to cancel noises up to 80 to 90 decibels when previously they were at 60 to 70 decibels.”
The most advanced noise cancellation technology uses multiple microphones to capture the optimum voice signal from the agent while effectively drowning out the ambient noise, using an advanced AI algorithm that preserves the natural tone and voice of the agent while doing so, Baliga explains.
The Inbound Issue
Contact center noise cancellation technology has evolved to the point where it is relatively successful at suppressing or eliminating unwanted noise in the contact center itself, commonly referred to as outbound noise, which is generated by the agents’ surroundings. It also needs to address the noise that comes through from the customer end of the conversation, commonly referred to as inbound noise.
Inbound noise makes it difficult for agents to comprehend customer inquiries and requests, muddying the clarity of the customer’s voice and leading to misunderstandings and longer call-handling times.
Inbound noise cancellation technology focuses on the impact of network conditions as well as the various codecs applied to the audio stream by communications systems. With it, advanced AI noise cancellation models are trained to dynamically adapt to varying network latency, jitter, and packet loss, ensuring optimal performance regardless of the communication platform or network infrastructure. Models are also trained to accommodate both mobile and landline audio signals and can even optimize noise cancellation algorithms for specific bandwidth constraints, ensuring consistent audio quality across devices and communication channels. And CPU-efficient models can cater to thin client environments, where computational resources might be limited, without compromising on performance or accuracy.
However, not all vendors offer top-level noise reduction technologies, and some of those with advanced technology don’t have solutions designed specifically for contact centers, cautions Dave Hoekstra, product evangelist at Calabrio, a contact center workforce optimization and analytics solutions provider. “While some have incorporated this technology, others have yet to do so. A key challenge is that speakers don’t receive feedback on what’s being canceled out, which can lead to skepticism about the technology, even though it works effectively.”
Testing Is Essential
Experts recommend testing any noise cancellation solution before installing it.
“With a variety of options available, choosing the right solution can be the difference between a professional customer interaction and a subpar experience,” Hoekstra says. AI-based noise cancellation technology is more widespread than many realize, and most users likely don’t even notice it’s working—apologizing for noises that the other person never hears.
To implement noise cancellation technology, contact center operators need to assess their communication methods and infrastructure to find something that integrates well with them, Miller says. “There is for sure no one-size-fits-all in this sense. Technology doesn’t always like to play well, so it may be a trial and error.
“There may be a need to upgrade software or invest in additional AI-specific tools to offer these advanced solutions,” Miller adds.
Communication solutions like Cisco and Avaya now also offer AI-powered noise-cancellation as part of their contact center solutions, Miller notes. “There are also many startups and newer companies that are focused on AI and NLP technologies.”
Baliga also urges potential contact center noise cancellation technology users to research vendors carefully and to fully assess their own IT environments. “The hardest steps will be training your existing staff on the new technology and integrating with your existing systems. This has been made easier with many solutions being offered as a software application or [software development kit]. This reduces any bumps in the road during the transition as they don’t have difficult pre-existing hardware requirements.”
Phillip Britt is a freelance writer based in the Chicago area. He can be reached at spenterprises1@comcast.net.