Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Help with listening tests (Read 14323 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Help with listening tests

Hi everyone, I could use some help with my listening tests.  My problem is that I can't hear the difference between a FLAC file and a 160kb/s (constant bitrate) MP3    (the album in question here is Radiohead's Kid A, maybe something less synthetic would be better?).  I have what I feel is a decent setup, nothing spectacular though (see below if you're curious, maybe it'll help).  My listening environment isn't exactly quiet though, so maybe that has an effect on things (fans in the computer are somewhat loud, fridge can be noisy, and the air conditioning is ridiculously loud...seriously, it wakes me up at night sometimes).

Anyway, back on topic...what exactly should I be listening for in my attempts to differentiate between these formats?  What is the extra bitrate of the FLAC file getting me?  (Sorry for the long first post, by the way; I've tried to be as informative as possible.)

Speakers:
pair of Paradigm Mini Monitor v5 bookshelf speakers (on 36" stands, approx 6' apart...none of that "throw them on the desk" stuff for me any more    )
    -amped with homemade amp (Google "gainclone" if you're curious...based off LM3875 op-amps, ~56-60W/channel RMS)
Hsu STF-2 10" sub (Low-pass at 85Hz, still working on high-pass to bookshelfs)

Input:
M-Audio Revolution 5.1 sound card (sub hooked up to LFE channel, so not using line-level inputs on it)
media players:
  Windows: Foobar2000 with ASIO output
  Linux: Amarok (dmix reconfigured so it doesn't resample to 48kHz)

Help with listening tests

Reply #1
Feel blessed that you don't hear the differences. Honestly, just keep the FLACs, make your MP3s, and call it an afternoon.

However if you want to know what to listen for: start with very low bitrate MP3s and work your way up to ~160kbps while performing ABX tests (FooBar2000 is good for that).

Me, I'm happy with Vorbis at Q2 (~92kbps).
OP can't edit initial post when a solution is determined  :'-(

Help with listening tests

Reply #2
Hi everyone, I could use some help with my listening tests.  My problem is that I can't hear the difference between a FLAC file and a 160kb/s (constant bitrate) MP3  ...

I wouldn't call this a problem. And you don't have to blame your ears or equipment: mp3 @ 160 kbps is expected to be transparent usually.

If you are really out for finding differences you must go for special kinds of music where mp3 encoders usually have weaknesses at 160 kbps. harpsichord music belongs to this (try harp40_1 which you can find on this forum). Impulses as with percussion instruments or electronic music are also often a problem (try castanets as a percussion sample and eig as an electronic sample). Issues from these fields are rather easily audible, but even then don't expect extremely annoying results from a good encoder @ 160 kbps.
lame3995o -Q1.7 --lowpass 17

Help with listening tests

Reply #3
Feel blessed that you don't hear the differences. Honestly, just keep the FLACs, make your MP3s, and call it an afternoon.


as much sense as that makes, if i were in the position of the OP i think my ego/pride/curiosity just couldn't handle that.  as someone who cares a lot about music and quality, it would feel like a "failure" of sorts and i'd be trying my darnedest to abx the difference!

i've found with any compression the easiest place to tell is the cymbals.  the mixture of high frequencies with the gradual decay.  try and find a few tracks where the cymbals are loud in the mix, makes it easier.

Help with listening tests

Reply #4

Feel blessed that you don't hear the differences. Honestly, just keep the FLACs, make your MP3s, and call it an afternoon.


as much sense as that makes, if i were in the position of the OP i think my ego/pride/curiosity just couldn't handle that.  as someone who cares a lot about music and quality, it would feel like a "failure" of sorts and i'd be trying my darnedest to abx the difference!

i've found with any compression the easiest place to tell is the cymbals.  the mixture of high frequencies with the gradual decay.  try and find a few tracks where the cymbals are loud in the mix, makes it easier.


Ha, that actually sums up my thoughts pretty well about that.  I agree there is a sort of determination on my part to try and discern a difference.  I can definitely hear WMA files' poor quality (especially considering most of the ones I've paid attention to are below 128kbps), so I assumed the huge difference in bitrate here would have amounted to a more sizable difference in sound.

Thanks to everyone for the advice so far; it sounds like percussion is the first thing to be affected by decreases in bitrate, which the Radiohead album doesn't have much of (its tracks are mostly slow and spacey, to throw in some subjective terms).  Time to make more mp3's for comparison....maybe I'll go with something acoustic, and with a lower bitrate (maybe the original CD track vs. a 92kbps MP3).

Help with listening tests

Reply #5
I suggest that you better try MP4 or OGG at ~96kbps. It is expected at that bitrate that it has better quality than mp3.

Help with listening tests

Reply #6
What I found on normal music is that 128k vbr sounds transparent on parts - esp loud music. With quite music - with solo vocal, solo guitar etc there will be some ringing and other artifact that can become annoying on some music esp with CBR encoding. Going to V4 reduces it a lot and transparency is reached on many samples. At V3 I cannot abx normal music and even some problem clips for V5 and V4 gave me trouble. Also a large percentage of problem clips for V3 are abxable @ V2 in my experience.

If the 128k tests are representative of  non-killer stuff , V3 (and maybe V4) will be as transparent as the old benchmark -- preset standard, extreme, insane to many people.

Help with listening tests

Reply #7
The stuff I find needs higher bit rates is a lot of metal, lots of cymbal crashes and stuff....

Help with listening tests

Reply #8
The stuff I find needs higher bit rates is a lot of metal, lots of cymbal crashes and stuff....


I agree that metal has quite a lot going on.  I still use 128kbps VBR iTunes AAC (I use iTunes AAC as Nero AAC isn't natively compatible with my Xbox 360) as it is transparent to my ears especially since I listen to music 70% of the time in portable environments, the rest of the time I listen to music while surfing the internet or doing homework so I am not really concentrating on it.

I wouldn't be too worried with trying to differentiate between high bitrate lossy files and lossless files.  I wouldn't be able to fit much on my 16GB iPod touch if I went with -V 2 with Lame or 192kbps VBR with Nero/iTunes AAC especially since I have about 8GB worth of video content on it.

Help with listening tests

Reply #9
maybe I'll go with something acoustic,

i have found the quieter and less complex the music the harder it is to abx quality differences.  i have also found that when flac compressing recordings solo acoustic music always compresses a lot higher than full band music - meaning there is less to capture in acoustic and that 128 or 160 kbps is going to get a lot closer to the uncompressed sound with less work.

so i would suggest the opposite, as was mentioned metal and hard rock has lots of cymbals and complex sounds, it is the easiest i think to pick up.

but yes starting low and working up on the same samples makes it easier too, as the problems / artifacts are the pretty much the same, just more or less severe at different bitrates.

if you like radiohead try kid a, maybe national anthem?  complex tracks like that.

Help with listening tests

Reply #10
I find solo intros easier. V5 will do ok on loud music but quite tracks with lead strings or vocals have ringing to some extent. Maybe the beginner listener always looks for the 'hard complex ' stuff..

Help with listening tests

Reply #11
LAME 128 kbps CBR is quite transparent for a lot of music. I find "impossible" to ABX guitar-based music. The easiest is electronic music, synths, cymbals galore. You shouldn't get that surprised. The forum listeners already came to a consensus as for not doing any 128kbps listening tests anymore because they don't see the point anymore and they were planning going down to 96kbps. (someone please cite that thread in which goorubulez said this).

So you shouldn't be surprise with Radiohead 128/kbps.
I find ABX only useful to test killer samples, but as for "normal" listening, it's pretty difficult to spot differences on non-killer samples.

Help with listening tests

Reply #12
Be glad and enjoy your compressed music as it is now.  Ear training will only change how you hear the music.  Instead of just enjoying it for its own sake and its aesthetics, you'll become a critic and the common flaws will bother you enough to be annoying.  Leave it alone and simply appreciate your collection at 10% of its original size

Help with listening tests

Reply #13
Instead of just enjoying it for its own sake and its aesthetics, you'll become a critic and the common flaws will bother you enough to be annoying.

It's even worse than that: if you have a hint of OCD, you'll be *looking* for flaws. My ABX days totally ruined the enjoyment I got from listening to music, during that period. Now I'm back to normal (I don't care) and I'm able to enjoy lossy music again

 

Help with listening tests

Reply #14
I agree, Mp3 sounds good on most popular music today. I can sometimes here a difference,but only if I really look into it. Mostly this is is with Jazz artists especially female vocals have a bit more sibilance and the bass seems less full. Modern Cool by Patricia Barber comes to mind...