GSMArena feature labs: The tests
When we designed our original speaker loudness test (now discontinued), the goal was to gauge how loud your phone will ring when you get a call or how well the phone is geared towards use in speakerphone mode. That's why we simply measured the peak loudness of the phone’s speakers in decibel while playing three different audio samples (a classic phone ringing, some pink noise, and then human speech). It was great if you wanted to know how loud this particular phone was but it didn't give you any indication of how well it sounded.
With the advance of smartphones, nowadays phone usage leans more and more towards multimedia consumption - people gaming, listening to music and watching movies on their phones' speakers, so we decided that measuring the sound quality was equally important too. Also, we figured out how to measure the average loudness instead of the peak one, making the loudness score more indicative of the experience you will get with multimedia rather than when your phone's merely blasting high-pitched tones.
Our test includes playing back a few select tracks in an environment with controlled acoustic properties and recording that playback with an audio recorder.
The first benefit of our new test is that you can now listen to the recordings by all tested phones. This way you don't have to rely solely on our subjective commentary about what they sound like (we recommend using headphones for playback).
But you don't have to trust just your hearing on this. Our audio sample set also includes a specialized audio track that sweeps the whole frequency range. We analyze the recording of this track by specialized software that tells us how well the phone's speaker(s) reproduces the different frequencies.
A look on the frequency response chart will easily tell you whether one phone reproduced the bass, treble, and mid frequencies better than another phone. (Tip: The closer to the flat line of 0db, the better.)
And finally, based on the recordings we've made, we can analyze the average loudness of the speaker output, which gets us a loudness score. And since we now measure the weighted average loudness of a phone and not the peak one, this loudness score is not affected by peak bursts of high tones which are irrelevant for multimedia consumption.
LUFS stands for "Loudness Units relative to Full Scale" - it's an industry-standard unit used in audio normalization for broadcast TV as well as streaming music and video.
You should keep in mind two things when reading those. The LUFS unit represents a relative measurement - the number is derived when compared to a particular baseline. So unlike decibels, you can't compare them to the sound of a TV or the roar of an airplane. The number here will only serve the purpose of comparing one phone to the other in our particular test.
Note that higher is better and the numbers are negative, meaning that the closer the LUFS reading is to zero, the louder the sound. To make it easier for you, we’ve also assigned ratings to each phone, ranging from Poor to Excellent. These are not directly comparable to the ratings from our old loudness test since, as we already explained, they measure average rather than peak loudness.
Finally, you can use our new tool to compare any number of phones we've subjected to the new test. We've already retested a bunch of phones we had available, but going forward the new test will become a part of our regular review routine so there will be plenty to choose from.
How to use the new widget
You will find our new Speakerphone test widget embedded in its corresponding section in our reviews. It offers key info of how loud the phone is (measured in LUFS plus our rating) and it shows the frequency response chart, allowing you to instantly spot phones that have weak bass or whose output in the high frequencies is too strong, making the audio harsh.
The widget becomes even more useful when you add one or more phones to compare (tap Add to comparison and use the search). You can compare the frequency response graphs to get a feel for the audio. The chart goes from +30dB to -30dB, but keep in mind that -30dB means 1,000 times quieter (and +30dB is 1,000 times louder).
An ideal speaker would produce a flat line at 0, any deviations above or below the zero line are flaws. When comparing two phones, what you need to look at is how close the lines are to zero and also how flat they are - any peaks or valleys are audible distortions of the sound.
Better yet, you can hit the Play button and hear the difference in sound yourself. We recommend that you use headphones for this unless you have good quality speakers hooked up to your PC (if you’re reading the results on your phone, definitely use headphones).
We’ve preselected the first phone in the example below. The Asus ROG Phone II in Outdoor mode is currently the loudest phone in our database (and we think it will keep that title for quite a while). However, add the iPhone 11 Pro to the comparison, and you’ll instantly see and hear the difference in bass and that the ROG scarifies bass frequency reproduction in the pursuit of loudness.
In fact, that was another issue with our old test – our ringtone sample leaned heavily on high-frequency sound. Some phones had speakers that produced a tinny sound that were better suited to produce these high-pitched sounds. This accounted for a higher decibel rating in our old peak loudness test but subjectively, the sound was harsh and sometimes unpleasant even. Our new test shows more nuanced results, separating the practical (hearing a notification) from the entertainment side of things.
Use the Playback controls to listen to the phone sample recordings (best use headphones). We measure the average loudness of the speakers in LUFS. A lower absolute value means a louder sound. A look at the frequency response chart will tell you how far off the ideal "0db" flat line is the reproduction of the bass, treble, and mid frequencies. You can add more phones to compare how they differ. The scores and ratings are not comparable with our older loudspeaker test. Learn more about how we test here.
appreciate the new loudspeaker test reasoning but would really welcome comments on the in call quality and loudness of the earpiece because other than multimedia, calls are the primary use of a phone
- 03 Apr 2020
Do you do endurance tests with several loops on the Benchmarks to see if the phone has throttling? How are your benchmark scores calculated? Is it an average of several loops? if so how many? Thanks
- 01 Apr 2020
Please add 4G Talktime To the table and Some New Phones in the Table Like Realme 5i, Samsung Galaxy M31, Realme X50 pro, IQOO 3, Poco X2, Redmi K30 and Redmi K30 Pro
- 04 Mar 2020