オンラインマイクテスト

The meter shows RMS loudness — the root mean square of the audio signal over a short window — converted to a decibel scale and mapped onto 0 to 100. That is a mouthful, but the reason matters: a simple peak reading jumps on every door slam and makes a silent room look like 30 per cent, which turns a meter into decoration.

Human speech at a normal conversational distance, on a decent microphone with automatic gain enabled, lands somewhere between 30 and 70 per cent on this scale. A whisper or a microphone two metres away sits under 15. A microphone with its input gain set too high, or a headset boom resting against your lip, pushes past 90 and clips.

Clipping is worse than it sounds, because it is not a volume problem — it is a shape problem. Once the waveform hits the ceiling, the tops of it are cut off, and no amount of turning the volume down at the other end restores them. Consonants are the first casualty, which is why a clipped voice sounds loud and unintelligible at the same time.

Distance does more than any setting. A microphone at 15 to 20 centimetres picks up your voice far louder than the room; the same microphone at a metre picks up roughly equal amounts of both, which is what makes a laptop microphone in a hard-surfaced room sound hollow. Halving the distance is worth more than any noise suppression algorithm.

Position matters almost as much as distance. A boom microphone belongs at the corner of your mouth, not in front of it — directly in front puts every plosive consonant straight into the capsule, which is the popping you hear on p and b sounds.

Rooms with parallel hard surfaces produce a flutter echo that no algorithm removes cleanly. A curtain, a rug, a bookshelf or simply not sitting in the middle of the room makes an audible difference. Speaking towards a soft surface rather than a bare wall helps more than it sounds like it should.

Finally, wired beats wireless for reliability, and a dedicated headset beats a laptop's built-in array for everything except convenience. Bluetooth headsets in particular switch to a much lower-quality codec the moment their microphone is opened, which is why your music-quality headphones sound like a telephone during a call. That is a limitation of the Bluetooth profile, not of your headset.

A browser microphone test and a real telephone call have one important difference: the phone network is narrowband. A traditional call carries roughly 300 Hz to 3.4 kHz, which is a fraction of what your microphone captures and a fraction of what a video meeting transmits. Voice sounds thinner on a phone call for that reason alone, and no microphone fixes it.

What that means practically is that a microphone which sounds merely acceptable in a browser test can sound poor on a phone call, because the frequencies that were carrying the clarity are the ones the network removes. Clear articulation and a close microphone matter more on a phone call than on a video meeting, not less.

Telvio places calls over an internet connection to ordinary phone numbers, so it sits between the two: the leg from your device to our network can carry wideband audio, and the leg from our network to the recipient's telephone is limited by whatever their operator supports. Calling somebody's mobile in a country whose network supports wideband voice will sound noticeably better than calling a landline on legacy equipment, and there is nothing either end can do about the difference.