TL;DR
I created a free blind‑test tool to see if lossless audio is actually audible:
https://georgebogdanov.com/tools/blind-audio-test
Key advantages of this tool compared to existing alternatives:
It allows you to use your own files – drag in two versions of the same track to test yourself if you can hear the difference.
It’s browser-based, so you don’t need to download or install anything.
No file leaves the device; everything runs locally in the browser. Works offline after load.
If this gets traction, ABX‑style rounds are a realistic addition.
It should be no surprise that I love listening to music. However, I am not an audiophile at all. I’ve been mostly satisfied with $30 headphones most of my life – embarrassing to admit as a musician. For me, it was always all about the content of the music: idea, composition, emotion, story, melody, rhythm, and instrumental mastery.
This was until I recorded my first album, Burst of Being, when I decided to confront my ignorance by fully mixing and mastering it myself. In an attempt to make my album sound good, I learned to appreciate the nuance of the sonic world the hard way.
Blind Test
While some streaming services are promoting lossless audio as a great premium feature, blind studies show that most people wouldn’t hear it in typical conditions and equipment:
While I completely failed NPR’s classic audio quality quiz, unable to distinguish between 320kbps MP3 and a WAV, I could definitely hear the difference when my own track was smashed down to something like 56k.
This made me curious: What is the lowest MP3 bitrate at which I could hardly distinguish it from lossless audio with my own mixes? (spoiler: 192kbps)
To answer this question, I’ve vibe-coded a tool. After having so much fun using it, I realized it might be useful for others. So, after adding some bells and whistles, I made it available for free use:
https://georgebogdanov.com/tools/blind-audio-test
My guessing score with ‘Burst of Being’:
WAV / 128kbps: 10/10 (100%)
WAV / 192kbps: 7/10 (70%)
Note that if you are comparing two variants, a score of 5/10 (50%) is the same as a random chance, statistically speaking. Only a score above 50% can indicate that you are guessing correctly. This isn’t a lab study, though – just a practical check.
And in the wild, streaming quality can swing dramatically too – here’s what happened to my album…
From Ultra Lossless to Ultra Lossy
Streaming services can go the other way too, by compressing the heck out of your audio!
This is what I discovered when I first streamed my album from Spotify over a cellular internet connection. It turns out that my iPhone had the “Low Data Mode” setting on, which caused Spotify to lock into the “Data Saver” setting, resulting in music being streamed in the lowest quality.
I noticed it is because my ears are so used to hearing my own mixes in high quality – I know what they should sound like.
Just to give you an example. Here is how I used to hear the climax of my Merry-Go-Universe song (this is a 192kbps MP3):
And this is what it sounded like with “Data Saver” on Spotify (although I don’t know which compression they’ve used, I tried to replicate it as close as possible, which came down to a 56kbps MP3):
In my case, iOS “Low Data Mode” forced Spotify into “Data Saver”, and it took multiple steps, including clearing the Spotify cache, to restore normal quality.
My main problem with Spotify, however, is that they don’t even indicate which quality is currently being streamed. For example, I like how YouTube does it, showing a quality dropdown that defaults to an automatic option, and allowing you to change it.
What I learned
192kbps MP3 is sufficient for my current needs. Of course, there are other efficient formats, such as FLAC/ALAC, that provide practical lossless delivery alternatives to WAV, but I have left them out of scope. (I use FLAC in my music production pipeline.)
96kbps would be the lowest boundary for me – anything below sounds exponentially worse.
The lower the MP3 bitrate, the more dull, narrow, and compressed the sound becomes. At ~96 kbps and below, artifacts are often audible.
Generally, the lossless sound is clearer, wider, and crisper in the high end, and the low end is fuller/warmer. It is easier to judge by busier song sections, as MP3 tends to struggle more when a lot is going on across the spectrum. On the contrary, when there is just a single instrument like a solo piano line, it is very hard to hear the difference between lossless and lossy.
I initially went hunting for obvious MP3 artifacts and mistakenly blamed the ‘harsh’ top end on lossy compression, but it was the opposite: lossless preserves upper‑harmonic detail and transient clarity, while low‑bitrate MP3 can smear or blur the highs; with strings, the bow noise and micro‑textures come through much clearer in lossless, a cue that helped me identify it. Honestly, for pure enjoyment, there are moments I even prefer the softened MP3 take.
There are different MP3 compression modes: VBR, ABR, and CBR. VBR generally yields better perceived quality at the same average bitrate; however, I used CBR for stability and repeatability in tests.
Although most listeners will hear my music as lossy streams alongside other equally compressed tracks, I've noticed that some mixes survive compression better than others. I’d love to hunt for strong references – especially string‑heavy tracks – but I’m not yet keen on spending hours in Spotify’s lowest‑quality mode.
P.S.
As technical as this exercise can be, I couldn’t help but get emotional from music. As I listened to ‘Ancient Bristlecone’ during this blind test, it actually made me cry, and later ‘Merry-Go-Universe’ did the same. This is the first time it has happened to me on camera, so I thought it would make a nice clickbait 🤪
Enjoy the tool, and please share your results here in comments or with me directly!
Music in the video: Strange New World by Pur Pasteur
