Bluetooth speaker test — channels, sweep and bass

This page plays test signals through whatever speaker your browser is currently using: a left-only tone, a right-only tone, a variable tone between 20 Hz and 16 kHz, an eight-second sweep from 20 Hz to 20 kHz, a slow 20 to 200 Hz bass sweep, and a band of noise. Between them they answer three questions: are both channels arriving, does anything rattle, and how low does this speaker really go. Turn the volume down before you press anything.

The six buttons, and what each one is for

Every signal is generated in the page when you press the button. Nothing is downloaded, nothing is recorded, and this page never opens your microphone.

Button	What it plays	What you are listening for
Left only	Two seconds of tone in the left channel, silence in the right	Sound from the left driver, and on a true stereo rig only from it
Right only	The same tone, right channel only	A matched level from the other side
Test tone	Two seconds at whatever the slider says, 20 Hz to 16 kHz	A clean steady note. Buzz or fizz means something is loose
20 Hz to 20 kHz sweep	Eight seconds, exponential, so roughly equal time in every octave	Rattles, resonances, and any band that drops out
Bass sweep	Ten seconds from 20 Hz to 200 Hz, same exponential shape	The frequency where the note stops descending
Noise	Three seconds of broadband hiss	Cabinet buzz, and an A/B against a second speaker

The exponential shape matters. A sweep spending equal time on every hertz would spend most of its eight seconds above 5 kHz. Spacing it by octave gives the full sweep about eight tenths of a second per octave, and the bass sweep roughly three seconds per octave, slow enough to hear the note where a small speaker stops keeping up.

Start quiet, because a sweep at full volume is how small speakers die

Two volume controls are in play and they do not always agree. Most Bluetooth speakers keep their own level separate from the phone's, so a speaker sitting at maximum will take a signal your phone thinks is quiet. Check its control before the sweep, not after. Your ears deserve the same caution: a 12 kHz tone at party level tells you nothing, because you flinch before you can judge it.

Left, right, and the fact that most Bluetooth speakers are mono

If the left-only tone and the right-only tone both come out at the same level, that is almost certainly not a fault. Most portable Bluetooth speakers sum the two channels into a single driver, and the ones with two drivers usually sit them only a few centimetres apart, which is why the stereo image barely separates even when both channels are genuinely there. The buttons still earn their place on a mono speaker: they prove both halves of the signal are arriving, the thing that quietly fails on cheap cables and adapters.

The test earns more on real stereo systems. In a car it identifies which door the sound comes from and catches a speaker left miswired by an old repair. On a soundbar it catches a wrong input mode. On a stereo pair of Bluetooth speakers it tells you which box is left, and if the boxes also disagree in timing you have the out-of-sync problem a stereo pair creates rather than a channel problem.

Reading the full sweep by ear

Play the 20 Hz to 20 kHz sweep with the speaker on a hard table, then again with it in your hand. Four things are worth noting.

• A buzz that appears at one point and vanishes. Something is resonating. If it disappears when you pick the speaker up, the table was buzzing. If it follows the speaker, drag the tone slider there and hold the note while you press on the grille and the seams.
• A band that goes quiet. Real dropouts mid-range are rare and usually mean a dead driver in a two-way box. Small dips are normal and are as likely to be the room as the speaker.
• The point where the top end disappears. Plenty of adults hear nothing above the mid-teens of kilohertz, which is ordinary age-related hearing loss rather than a speaker fault. The tone slider stops at 16 kHz for that reason.
• Anything that changes between runs. Many Bluetooth speakers apply their own limiting, so the same sweep can behave differently at two volumes. That is the speaker's processing, not an intermittent fault.

The bass sweep, and where a small speaker gives up

Start the 20 to 200 Hz sweep and follow the note down. There is usually a clear moment where the pitch stops descending and the sound turns into chuffing air from the port, or a thin whine an octave above where the note should be. That point is the speaker's honest bottom, and on a pocket-sized Bluetooth box it often sits in the low hundreds of hertz rather than at the figure on the label.

That number changes what to expect from a voice through the same speaker. The app's Deep preset is a 480 Hz lowpass with a 160 Hz shelf lifted 6 dB, and Warm puts an 8 dB shelf at 240 Hz, so both put energy in a region a small speaker may not reproduce at all. You can hear the difference with the eight voice presets and their frequency-response curves, then pick the one that suits the speaker you own.

Do not use the bass sweep to compare two speakers by which sounds bigger. The louder one wins that comparison whatever it reproduces. Match levels by ear on speech, put both in the same spot, then sweep each.

What a browser test cannot tell you

This tool only plays. It has no return path, which makes it different in kind from the latency test on this site, which plays a chirp, records it back through the microphone and cross-correlates the two into a number. Here the browser has no idea whether your speaker made any sound at all. Every judgement is yours, made by ear, and that has hard limits.

• This is not a frequency response measurement. Getting one needs a calibrated measurement microphone, and even then the result describes that speaker in that room in that position. Move a speaker a metre and the bass changes by more than most speakers in a price bracket differ from each other. What you hear is speaker plus room plus your own hearing, and listening cannot separate them.
• There is no level reading. No decibel figure, no sensitivity, no distortion percentage. The phone SPL app you might reach for is not calibrated either.
• The signal is not delivered untouched. Over Bluetooth it is encoded before your speaker sees it, and then the speaker's own processing takes a turn. An iPhone sends AAC and nothing else. An Android phone since 8.0 can negotiate SBC, AAC, aptX, aptX HD or LDAC, though only LDAC's encoder ships in AOSP: aptX and aptX HD are proprietary Qualcomm code each maker licenses, so whether your handset offers them depends on the handset. Android also exposes a Bluetooth audio codec picker in Developer options, which iOS has no equivalent of; switching codec shifts latency somewhat rather than dramatically, and it can change what you hear on this test. You are testing a whole chain and hearing its worst link. The codec guide covers which one your phone uses.
• Silence at 20 Hz is not a result. Almost nothing portable produces the bottom octave, and the sweep spends under a second there.

What it is good for is the class of fault that ruins an event: a dead channel, a driver that has stopped working, a grille buzzing, a speaker with no usable low end. Those are audible in ninety seconds, and they are most of what goes wrong.

Once the speaker passes, test the rest of the chain

A working speaker is half of a microphone setup. If the plan is to put your voice through it, check the input end with the microphone test, then listen to the whole path with the live mic monitor before anyone is watching. If your voice comes back late, that lag belongs to the Bluetooth link rather than to the speaker you just tested, and the guide on where Bluetooth microphone delay comes from explains which part is fixed and which part you can work around. Singers feel it worst, the subject of the karaoke delay guide. The app that does the sending, Bluetooth mic: Mic to Speaker, is free on both stores, so an Android phone and an iPhone can both drive the speaker you just tested.

The other failure a speaker test will not catch is the howl that starts when a live microphone hears its own output, a matter of geometry and gain covered in the guide on stopping feedback. If this is event preparation, work through the pre-party checklist the day before, not in the fifteen minutes while people are arriving.