• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

I found an abx test on a website

jannek

Active Member
Joined
Mar 21, 2020
Messages
233
Likes
288
Location
Southern Germany
After doing the 320 vs lossless test with maximum frustrating 100% guessing (and about 50% right) I changed to the easiest one with 96kBit/s Lame MP3 vs. lossless to verify the files are different at all. This was a lot easier, at least the first four tracks. 98% correct. Haven't had the time since to check higher bitrates yet...
 
OP
I

ILikeMusicToo

Member
Joined
Mar 1, 2021
Messages
7
Likes
11
Location
West Palm Beach, FL
I'm going to work my way up from the lowest quality and see where the test tells me I'm scoring at the level of chance. I just did the 96k mp3 vs lossless and, thank the audio gods, I passed with 80% correctly identified. This was on my laptop with a Dragonfly Black USB dac and a set of Thinksound ON2 on ear headphones.

The Killers
100%
(p >= 0.020)
James Blake
100%
(p >= 0.020)
Daft Punk
60%
(p >= 0.020)
The Eagles
100%
(p >= 0.020)
Dixie Chicks
40%
(p >= 0.020)

The higher quality samples sounded shade more "open" and the lower quality tracks had a bit of "swishing" in complex passages, which is a typical signature of low bitrate, I believe. But dang, it was close, and you can see on two of the tracks (Daft Punk and Dixie Chicks) I am already at the level of chance.
 

taner

Active Member
Joined
Jun 13, 2020
Messages
199
Likes
136
I didn't take it seriously on first (The Killers) after got %40 I tried my best on next (James Blake) and got %70. James Blake was easy. On the lossy one, there is not much ambience/echo as lossless one. I will continue later.
 

paulraphael

Active Member
Joined
Dec 18, 2020
Messages
262
Likes
367
Location
Brooklyn, NY
Has anyone tried creating a deliberately messed up eq curve (maybe something that creates comb filtering) to see if it reveals compression artifacts? Would be a kind of interesting experiment.
 

j_j

Major Contributor
Audio Luminary
Technical Expert
Joined
Oct 10, 2017
Messages
2,267
Likes
4,759
Location
My kitchen or my listening room.
Has anyone tried creating a deliberately messed up eq curve (maybe something that creates comb filtering) to see if it reveals compression artifacts? Would be a kind of interesting experiment.
Well, if you heavily shape the OUTPUT of a codec, yes, that may get you trouble. Shaping the input should, with a decent encoder, be ok.
 

paulraphael

Active Member
Joined
Dec 18, 2020
Messages
262
Likes
367
Location
Brooklyn, NY
Well, if you heavily shape the OUTPUT of a codec, yes, that may get you trouble. Shaping the input should, with a decent encoder, be ok.

I was thinking of it more as an experiment to reveal artifacts, and to get a sense of how uneven curves can ruin the masking effect.
 

j_j

Major Contributor
Audio Luminary
Technical Expert
Joined
Oct 10, 2017
Messages
2,267
Likes
4,759
Location
My kitchen or my listening room.
I was thinking of it more as an experiment to reveal artifacts, and to get a sense of how uneven curves can ruin the masking effect.

Well, done on the output, that's exactly what happens. On the input, the perceptual model OUGHT to cope. Famous last words, maybe, but it should work.
 

Jostein

New Member
Joined
Jan 13, 2022
Messages
1
Likes
0
Hi everyone,
I have wondered, as I'm sure many of you did, if my system could play music well enough to let me distinguish between lossy and lossless streaming music. I'd tried comparing my FLAC CD rips to the same album streaming on Spotify (I don't have Tidal/Qobuz) and honestly couldn't hear the difference, but wasn't sure because I couldn't switch instantly between the sources.

I found this website (http://abx.digitalfeed.net/#) that has an ABX test with five music samples, played five times each, covering a range of music types. Sadly I couldn't tell them apart, scoring 52% and 48%. I intend to try different speakers, tweaks, speaker positions to see if I can find some sonic signature that stands out.

I just noticed that if you dig around they have other tests: like can you identify the "X" when one candidate is lossless and the other is a 96k encoded mp3! I mean...surely...right?

If someone scores much higher than 50% PLEASE reply here and share details of your system!

I'm running an Outlaw RR2160 amp, Mission 774 tower speakers (they were about $US 1200 twenty years ago), generic monoprice cable. I used a laptop as my source, once with USB into the Outlaw using its internal DAC, and a second test using my Dragonfly Black to feed the Outlaw's aux socket. Both times I played the music around 75 db, peaking at 80db.
I tried on headset once now (shortest version), and got 72% correct.
 
Top Bottom