Speakerphone and Audio

The Nexus 4 has a small vertical notch cut out of the back glass for its speakerphone port. The unfortunate part is that when the Nexus 4 is laid display-up like I always place phones for testing under our digital sound level datalogger, it is quite muted since there is no gap in the cavity for sound to escape through.

Speakerphone Volume - 3 inches Away

With the phone raised, however, the Nexus 4 turns out to be decently loud, which matches my subjective impressions using the device for Google Navigation over the past few days.

Noise Suppression

The Nexus 4 has a pair of microphones for noise suppression both when on calls, and also for the increasingly important task of reducing noise on ASR (Automatic Speech Recognition) workloads like Google Now. I believe the Nexus 4 is using Qualcomm’s Fluence for this task, which is an adaptive beamformer system.

Google Nexus 4 - Noise Rejection by AnandTech

To test its efficacy, I turned to the industry standard babble track and ramped volume in front of a pair of speakers to 94 dBA (very loud) and then back down while recording the mobile-terminated end of the call on my PC. I should note that when I run these tests I always originate and terminate the call on the same mobile operator (in this case T-Mobile) if possible.

The Nexus 4 does a pretty decent job at canceling noise on my test call. The Galaxy Nexus noise rejection performance quite honestly never was that spectacular, and getting better noise filtering is going to be an increasingly important part of the speech recognition battle on these platforms.

Audio

Inside the Nexus 4 is a Qualcomm WCD9310 audio codec, which we’ve seen in other devices like the MSM8960-based Galaxy S 3s and a few other phones. Measuring sound quality is probably the number one requested addition to our reviews, and still is a rather nebulous thing to measure at times. For this I worked together with the ever-awesome François Simond (@supercurio) to measure sound quality on the Nexus 4 using RMAA on my desktop equipped with an ASUS Xonar Xense sound card.

Subjectively the Nexus 4 doesn't sound terrible to my ears on a pair of SE535s and listening to music at half volume or less. Objectively however the results are less than awesome thanks to a combination of things. First, audioflinger is set to 48 kHz which results in software resampling causing artifacts for 41.1 kHz source material. Second, there appears to be different modes that the Nexus 4 switches into depending on your volume level, and the frequency response plots show these different plots at the number of different volume levels we tested. We're going to update with some thoughts from Francois about the Nexus 4 soon, for now I think the Nexus 4 sounds ok at least when it comes to the most glaring of things — I couldn't detect any background hiss or whine as the SoC changes states, which is a huge percentage of what I normally wind up hearing on smartphones. 

Cellular, Wi-Fi, GNSS Conclusions and Final Thoughts
POST A COMMENT

188 Comments

View All Comments

  • numberoneoppa - Tuesday, November 13, 2012 - link

    Lovely review, Brian, as usual, but I have a question. Forgive me if I just glanced over this, but have you run your benchmarks using the newest OTA which rolled out over the last day or are these running the same software as those posted earlier in the month? In any case, it would be interesting to see if there are any changes in the XML file which seems to specify temperature shelves. Reply
  • Jorange - Tuesday, November 13, 2012 - link

    Hi Brian,
    Can you please confirm that all GL Benchmark tests are run serially for all devices?, as this places them under a greater thermal load, then running the benches individually.
    Reply
  • Brian Klug - Tuesday, November 13, 2012 - link

    I do whenever possible, always. The issue is what happens when devices crash out before a full run (Optimus G) which then forces my hand to do this piece-by-piece kind of testing and reassemble a whole run.

    -Brian
    Reply
  • MadMan007 - Tuesday, November 13, 2012 - link

    You should note which devices have 'assembled scores' versus straight runs in your charts then. An asterik would do. Reply
  • xTRICKYxx - Tuesday, November 13, 2012 - link

    By far the best Nexus 4 review; bar none.
    Kernel developers will easily raise the thermal throttling limit to a much higher temperature, but it is obvious there is a temperature issue.
    Reply
  • suhan - Tuesday, November 13, 2012 - link

    why is it that you don't use geekbench to measure the actual performance of the device? Reply
  • tuxRoller - Tuesday, November 13, 2012 - link

    Because it is closed source thus we dont know exactly what paths it takes (meaning it could disproportionately harm one particular device over another)?
    Hell, do we need even know what compiler they use? Knowing that and the flags used would certainly be very useful.
    Reply
  • HisDivineOrder - Tuesday, November 13, 2012 - link

    ...this site used to be about custom-built computers?

    We'd get driver reviews, we'd get hardware other than video cards, motherboards, and CPU's reviewed, we'd get the skinny on new custom PC hardware that was coming...

    I'm waxing nostalgic here. I miss those days, I think. These smartphone reviews are just beginning to bore me to tears. Maybe I'm just old, but I miss the good ol' days. Back when phones were something we just lived with rather than something people seem to obsess over now.

    Back when having the latest and greatest in computing was more about smart hardware selection and less about going to the store, slapping down your cc, and saying, "Gimme your best, barkeep!"
    Reply
  • Medikit - Tuesday, November 13, 2012 - link

    I remember those days really well. Mobile computing is the future of the industry. Furthermore Anandtech still reviews computer cases, HTPCs, and the latest, greatest hardware. I like to think of it as the best of both worlds. Reply
  • Alexo - Thursday, November 15, 2012 - link

    > Mobile computing is the future of the industry.

    "Mobile computing" will not get truly mobile until we get decent battery life.
    Meanwhile, AnandTech keeps lauding devices that have more computing power than most of us ever utilize at the expense of battery life (not to mention non-user-changeable batteries, throttling, etc.)
    Reply

Log in

Don't have an account? Sign up now