---
title: "ABX Blind Test Online: Can You Hear the Difference?"
description: "Free online ABX blind test, sample-accurate and level-matched. Six synthetic test pairs (THD, polarity, 14/16 kHz hearing) or drag in two files of your own."
url: "https://theaudiostuff.com/tools/abx-test/"
type: "website"
author: "Jakub Charkiewicz"
---

> Markdown rendering of https://theaudiostuff.com/tools/abx-test/
> Canonical URL to cite: https://theaudiostuff.com/tools/abx-test/
> Generated from the published page. Text and links only; see the HTML for layout.

Breadcrumb: [Home](https://theaudiostuff.com/) > [Tools](https://theaudiostuff.com/tools/) > ABX Blind Test

# ABX Blind Test: Can You Hear the Difference?

A free online ABX blind test - you hear A, you hear B, then X is one of them at random and you say which. Sample-accurate and double-blind, in your browser, with the playback position carrying across every switch - the closest thing to "is this thing actually different" you can do without a measurement rig. Try a built-in synthetic test in two clicks, or drag in your own pair.

**New to blind testing?** Start with a built-in test below - two clicks and you're comparing.

My tests

Completed tests save here automatically. Click any saved test to re-open its result and certificate.

Score - Take the test to see your number.

Could it be luck? - The p-value: probability your hits came from random guessing.

Confidence

Awaiting test Five levels: chance · suggestive · significant · strong · publishable.

## Pick what to compare {#setup-heading}

setup

## Round 1 of 10 {#test-heading}

live

**Loop a section**

Focus the test on a few specific seconds (a vocal phrase, a cymbal crash, a quiet decay). Position carries across A/B/X switches inside the loop window.

s

s

Quick set

Loop a short section (2-5 s) of the most revealing passage - cymbals, vocal sibilance, or a bass drop.

Is X the same as A or B?

**Pro tip:** switch *quickly* between X and one suspected reference. Your auditory memory is < 4 s; long pauses defeat the test. The position carries across switches so you can A→B→X seamlessly mid-bar.

## Test results {#result-heading}

done

## How ABX testing works {#how-it-works}

### The protocol

ABX is the gold standard for auditory discrimination. You get two known references (A and B) and an unknown X, randomly assigned to be A or B each round. Audition each as many times as you like, then commit an answer.

Sample-accurate switching matters. The position carries across A→B→X in this tool, so you swap mid-phrase. Your auditory echoic memory is roughly 3-4 seconds; if the switch resets the playhead, the gap erases the trace you were comparing to.

### The statistics

The p-value is the probability of getting at least your hit count from pure random guessing on a binomial distribution. The standard "this is real" threshold is p < 0.05.

For 10 rounds: 8 ≈ p 0.055 (borderline), 9 ≈ p 0.011 (significant), 10 ≈ p 0.001 (highly significant). Longer tests need higher absolute hit counts but lower percentage: 15/20 (75 %) is p 0.021, comfortably significant.

## What your hit rate actually proves {#ref-abx-title}

Binomial probability of guessing the listed score or better by pure chance. The bar for "I really heard a difference" is `p < 0.05` - anything below that and the guessing explanation runs out.

| Score | p-value | Verdict |
| --- | --- | --- |
| 5 / 10 | 0.623 | Pure guessing. No evidence of audible difference. |
| 6 / 10 | 0.377 | Inconclusive - could be a slight bias, could be luck. |
| 7 / 10 | 0.172 | Suggestive but not significant. Run a longer test. |
| 8 / 10 | 0.055 | On the edge of significance (p ≈ 0.05). |
| 9 / 10 | 0.011 | Significant - only 1.1 % chance of guessing. |
| 10 / 10 | 0.001 | Highly significant. You hear it. Go publish. |
| 13 / 16 | 0.011 | Equivalent confidence on a longer test. |
| 15 / 20 | 0.021 | Clean significant result on 20 rounds. |
| 17 / 20 | 0.001 | Highly significant. Beyond chance. |

## What a blind test can and cannot prove {#tprimer-title}

An ABX test gives you two known samples, A and B, then an unknown X that is one of them, and asks which. Repeat it enough times and the proportion you get right separates genuine discrimination from guessing - because guessing converges on 50%.

The statistics are simple binomial. Getting 8 of 10 right happens by chance about 5% of the time, and 9 of 10 about 1% of the time, so a high score is evidence of a real audible difference. The reasoning only runs one way, though: a null result means this difference was not audible to you under these conditions, not that no difference exists.

The value of blinding is that it removes everything except the sound. Knowing which is the expensive one, or which you just bought, reliably changes what people report - not because they are dishonest, but because expectation is part of perception.

### The ideas behind the controls {#tprimer-ideas}

- **Level matching**: The single most important control. A 0.5 dB difference is audible as "better", so any unmatched comparison tests level rather than the thing you meant.
- **Null result**: Not hearing a difference under test. Evidence about this comparison and this listener, never proof that two things are identical.
- **p-value**: The probability of scoring this well by chance. Around 0.01 at 9 of 10, which is why trial count matters as much as hit rate.
- **Expectation bias**: The measurable effect of knowing which is which. It is what blinding exists to remove, and it affects everyone.

### Common mistakes {#tprimer-mistakes}

- Stopping as soon as the score looks good. Deciding the trial count in advance is what keeps the statistics meaningful.
- Comparing at different levels. Anything louder tends to be picked as better, and the test then measures gain.
- Reading a null as proof of no difference, when it only bounds what was audible in these conditions.
- Running too few trials. Three correct guesses in a row happen by chance one time in eight.

### What this tool cannot tell you {#tprimer-limits}

- It is only as good as the level match. A fraction of a decibel between A and B is heard as "better" rather than "louder", and it will show up as a real-looking result.
- A pass says you told them apart on this material, on this chain, over this many trials. A fail is not proof the difference does not exist - it is a failure to demonstrate one at this sample size.

## ABX blind test FAQ. {#abx-faq-title}

How the test works, what counts as significant, and whether your audio files are uploaded anywhere.

1. ### What is an ABX blind test? {#abx-faq-what-is-an-abx-blind-test}
   An ABX test is a statistical method for proving you can hear the difference between two audio sources. You hear sample A, sample B, then a randomized sample X (either A or B) and decide which one X matches. Get enough trials right and the result is statistically significant; flip a coin and you will not.
2. ### How many ABX rounds do I need to be statistically significant? {#abx-faq-how-many-abx-rounds-do-i-need-to}
   The tool runs 10 rounds by default. Nine correct out of 10 clears the p < 0.05 bar (about a 1% chance of being a fluke) and ten out of ten reaches p 0.001. Eight out of 10 lands at p 0.055, right on the edge, so the tool calls that borderline rather than significant. Fewer rounds are less reliable; for a claim you want to defend, run 16 to 20.
3. ### Why does ABX testing matter for audiophile gear? {#abx-faq-why-does-abx-testing-matter-for-audiophile-gear}
   Because most "I can hear it" claims fall apart under blind, level-matched conditions. ABX cuts through expectation bias by hiding which sample is which. If you can pass an ABX test on two cables, two DACs, or two file formats, the difference is real; if you cannot, it likely was not.
4. ### Are my audio files uploaded anywhere? {#abx-faq-are-my-audio-files-uploaded-anywhere}
   No. The ABX test runs entirely in your browser using Web Audio AudioBuffers. Your files never leave your device, never hit our servers, never get logged. The Certificate of Auditory Transparency you can share is just an image. It contains the result, not the audio.
5. ### How do I level-match the two samples for a fair test? {#abx-faq-how-do-i-level-match-the-two-samples}
   The tool auto-applies RMS level matching when you load two files, so loudness differences do not bias the test. If your two clips have very different peak levels but similar RMS, the perceived loudness will be close. For maximum rigor, pre-normalize the files in a DAW before loading.

## Shopping the gear behind the math?

- [Best DACs](https://theaudiostuff.com/guides/best-dacs/)
- [DACs reviews](https://theaudiostuff.com/reviews/dacs/)
- [Denafrips Enyo 15th Anniversary Scored 9.1 out of 10](https://theaudiostuff.com/reviews/denafrips-enyo-15th/)
- [Henry Audio DA 256 Scored 8.2 out of 10](https://theaudiostuff.com/reviews/henry-audio-da-256/)
- [SMSL DL100 Scored 8.2 out of 10](https://theaudiostuff.com/reviews/smsl-dl100/)
- [Henry Audio DA 256 vs SMSL DO100 PRO](https://theaudiostuff.com/compare/henry-audio-da-256-vs-smsl-do100-pro/)
- [Henry Audio DA 256 vs SMSL RAW-MDA1](https://theaudiostuff.com/compare/henry-audio-da-256-vs-smsl-raw-mda1/)

## You might also use these tools {#rt-title}

- [Training **How good are my ears, really?** EQ Ear Training Game Train your ears to identify frequencies from 20Hz to 22kHz. Adaptive difficulty, five training modes, score tracking, and a shareable Golden Ear Rating.](https://theaudiostuff.com/tools/ear-training/)
- [Listening test **What sound do I actually like?** Sound Signature Quiz A blind A/B preference test over royalty-free music. Pick by ear between more or less bass, mids, treble, reverb, dynamics, even-harmonic warmth, and stereo width - then get your signature in plain English.](https://theaudiostuff.com/tools/sound-signature-finder/)
- [EQ **What does this EQ change sound like?** Online Graphic Equalizer A 10-band graphic EQ you can hear in real time over royalty-free music or your own file. Watch the response curve redraw live, then export presets for Equalizer APO, AutoEQ, Wavelet, or Poweramp.](https://theaudiostuff.com/tools/eq-simulator/)