The test checks whether you can tell compressed audio (MP3) from lossless (WAV). You listen, guess blind, and learn your bitrate threshold on your own gear.

ABX is a blind listening method. You do not know which file is playing and identify it by ear only. That removes bias like I know FLAC must sound better.

To find the bitrate where you actually hear a difference from lossless , on your headphones and device. Not what forums say, but what you hear. Also cures early-stage audiophilia. :)

Three versions of one track:

• A compressed MP3 (bitrate goes up as you progress)
• B the same track lossless (WAV)
• X either A or B. Guess which

All MP3 files are cut from one lossless master. We compare levels close to real streaming:

• 96 kbps internet radio, podcasts
• 128 kbps free streaming (Spotify Free, Deezer Free, etc.)
• 256 kbps YouTube Premium, SoundCloud Go+
• 320 kbps Spotify Premium, SoundCloud HQ
• Lossless WAV , reference for comparison

Lossless: WAV or FLAC , no data loss, like the master.
Compressed: MP3 , smaller file, the codec discards information. Lower bitrate means heavier compression.

Listen to A and B, then X. Choose:

• A if X matches A
• B if X matches B
• Skip if unsure , does not affect progress

Two correct answers in a row at one bitrate move you to the next level (96 → 128 → 256 → 320 kbps). The goal is the highest bitrate where you still tell MP3 from lossless.

Two wrong answers in a row at one level end the test. You learn your ceiling: the highest bitrate where you still heard a difference.

Yes. Skip is not counted as wrong and does not break your streak , you move to the next track at the same bitrate.

Until two correct in a row (next bitrate) or two wrong in a row (done). Usually a few minutes. You can retake anytime.

Yes. Many people cannot tell 320 kbps from lossless , especially in noisy rooms or on basic headphones. The test shows your real threshold, not how it should be in theory.

WAV files are heavier than MP3 , that is normal. Speed depends on your connection. A progress bar shows while the next track loads.

Yes. Every version of a track comes from one WAV master, then we run loudness normalization (EBU R128). Why: in ABX, the louder option almost always sounds better even when quality is the same. We level A and B before the files reach your browser so you compare compression, not volume.

We use a plain HTML5 player (<audio>), not Web Audio API and no extra JavaScript processing. Some browser ABX tools build graphs of GainNodes and filters , we deliberately avoid that: fewer layers between the file and your ears. True bit-perfect output from a browser is still impossible , audio goes through the OS stack , but we do not add an extra JS layer.

For our use case, quite a bit. We compare 96–320 kbps MP3 to WAV, not hi-res formats on a thousand-dollar DAC. MP3 artifacts at 96 kbps are huge , audible even on cheap headphones and through a normal Windows mixer. The test shows whether you hear streaming bitrate differences on your setup. It is a screening test, not a lab-grade ABX at ±0.1 dB.

Not our scenario. ABXtest compares MP3 bitrates to lossless WAV , what you meet in Spotify, YouTube and podcasts. 192 kHz vs 176.4 kHz or DSD tests are a different league and different tools (foobar2000, foo_abx).

Some say browser ABX is useless. This section is our reply, including to this AudioHUB article. ABXtest does not claim golden ear status on a reference chain worth tens of thousands. The job is different: quickly learn your bitrate threshold on a normal PC or phone, no install. For strict audiophile ABX with bit-perfect output, use foobar2000 + foo_abx , another tool for another job. For MP3 vs lossless on my headphones, a web test makes sense.