In the age of social media, texting, mobile e-commerce and video streaming it’s easy to overlook an experience hasn’t gotten better for smartphone users: talking on the phone.
 
Despite sophisticated smartphones and networks, many mobile users are not satisfied with call clarity. None of the 100-plus smartphones in Consumer Reports’ 2014 phone ratings earned better than a good score for voice quality. A large number of smartphones rated only as “fair.”
 
In larger part that is because device makers often shrink, flatten and cover speakers in plastic to improve their phones’ overall functionality. Even on a high-end smartphone that uses several microphones and noise-cancellation algorithms, a caller is not guaranteed clear sound, especially in noisy environments.
 
Change is happening slowly but there are promising new technologies are on the horizon. Start-up Cypher Corp. has built an artificial intelligence engine that analyzes the unique quality of the human voice that distinguishes it from other noises. Cypher could be bundled into new smartphones or deployed as a software update, says John Yoon, the company’s vice president of product design. Yoon says demos of the software are currently running on a variety of Android handsets—including LG, Samsung, Kyocera, Korea Telecom and Google Nexus—when it becomes publicly available later this year.
 
The three of the top service providers in the U.S.—Verizon Wireless, AT&T and T-Mobile—are addressing call quality with Voice Over Long Term Evolution (VoLTE), an add-on to their next-generation cellular systems that enables voice traffic to be carried over 4G LTE networks. More than 100 mobile operators in 75 countries have commercially launched High Definition (HD) voice services, according to the Global mobile Suppliers Association. Sprint is among the operators offering HD voice, a speech coder/decoder program—a codec—that extends the frequency range of audio signals transmitted over telephone lines. More than 70 operators are investing in VoLTE studies, trials or deployments and 11 have commercially launched HD voice services with VoLTE. Apple’s iPhone 6 and 6 Plus models, for example, support VoLTE.
 
HD voice depends on technologies found both on the network and in handsets—features like enhanced audio processing, multiple microphones, speakers and improved echo cancellation. Together, these features give mobile users clearer voice calls and a significant decrease in background noise such as street traffic and wind. Yet HD voice availability is limited and cannot work as intended unless both the caller and the receiver have an HD voice-capable device and are located within the HD voice coverage area, says Jeremy Green, principal analyst at London-based Machina Research.
 
To understand problems with call quality and how they are being addressed, Scientific American interviewed Jerry Gibson, professor of electrical and computer engineering at the University of California, Santa Barbara.
 
[An edited transcript of the interview follows.]
 
Why is cell phone call quality so bad?
It is primarily the service providers. A key point to note is that the base station/base station controller, now called eNodeB [for Evolved Node B] in LTE, is all-powerful. That is, eNodeB makes all of the decisions about how much bandwidth each handset gets no matter how good a channel connection a handset may have. Also, base station behaviors are not standardized—that is, no one really knows how they are making these decisions. They take into account how loaded the cell site is and how loaded adjacent cell sites are, plus other network data and other things when allocating bandwidth. A couple of rules providers appear to follow are: Don't drop a call in progress, and don't block any new calls if at all possible. This means that the base station allocates bandwidth conservatively, and thus the voice codec in the handset may operate at a lower than desirable rate.
 
What is the big focus for service providers?
One—video is king these days, so video sucks up lots of bandwidth. For video, the service providers do not want you to have to wait for buffering, particularly after you have started to view a video. Second, cell sites/base stations cost money, so deployments of new cells are not done lightly. The latter leads to poor RF [radio frequency] connectivity in different areas for different providers.
 
What's at stake if we continue to have terrible call quality?
I don't think the service providers feel voice quality is their main problem today. They want traffic, and video is big traffic compared to voice. Video is expected to dominate mobile data in the future so it is important to their business. The current question of service providers is how to collect money per bit of data. What they say is: “How do we monetize video?”
 
What are some of the more promising projects aimed at improving call quality?
For the latest generation of cellular, LTE, a new voice codec is being developed. It is designated enhanced voice services [EVS]. It will cover variable [wider] bandwidths so it will be better for music and mixed voice and music content. It has many new codec rates, plus better VoIP [voice over Internet protocol] factors such as packet loss concealment [used to used to mask the disruptive effects of lost or discarded data packets] and jitter buffer management. [“Jitter” refers to variations in the length of time to deliver data packets.] But the standard will take awhile to get into deployments everywhere by service providers.
 
When fully deployed, LTE with EVS will be a big improvement—if video traffic does not take all of the bandwidth wherever you are, causing eNodeB to only allocate your voice call a low rate. In some ways the approach of service providers is understandable. No one wants a call dropped and no one wants their calls to be blocked [unable to get through]. Plus, video is something everyone appears to want on their mobile device.