The Great MP3 Bitrate Experiment

What about gapless playback? There still no general standard for this in MP3.

This discussion feels rather akin to discussing the quality of Bluray encoding whilst ignoring the capabilities of your television. The best encoding or lossless storage in the world won’t help if your playback equipment is introducing all sorts of noise, clamping and/or artefacts all by itself. The same can be said of the choice of encoder and the material given to it - any resampling needs to make sure it avoids clipping, normalising, or any other common older tricks to reduce file size or you absolutely will create noticeable differences.

The correct assertion is to encode to the quality of your intended playback equipment & scenario, with the right encoder (configuration) for the job. That is why, as the linked Neil Young article states, encoding for recording is done at 24-bit - because in recording you need to fully capture the signal without guessing the 16-bit range in advance.

That said, this test is deliberately goading the tester to find differences, and as the same article highlights, the human brain is great at hearing things that aren’t really there when it’s looking for them.

I notice with interest that the audiophile discussions revolve entirely around the 20Hz - 20Khz range because that’s what the ear hears, whilst completely forgetting that we are not just our ears. Has anyone done any studies on the sub/super-sonic effects on the physical body and how that may or may not influence the listener? I could easily imagine that an orchestra might produce e.g. 4Hz waves that might affect the physically present listener that a filter would entirely remove. Of course, that might be a good thing (3Hz being known to induce resonance in the stomach)!

Regardless I’d certainly rather have a lossless file so that I can choose to encode it to my hearts content for the right situation, to the file format that my given player supports.

I think you didn’t touch on an important side of the issue here. While MP3 is the most popular format now in the future it may fall completely by the wayside and you might have a device that doesn’t support it or simply want the advantages of another format. By storing all of your music using a lossy codec and getting rid of the reference media you’ve lost your ability to re-encode it without reducing the quality further.

A lossless codec such as FLAC does take more storage space but it also allows you to re-encode to something else without any quality loss. Personally I’d much rather take the space hit to store my music as FLAC on my home server (Running out of space with 6TB would take a LOT of music). Encoding from FLAC to MP3 (or another format of choice) is fast enough on a modern system that I’ll simply re-encode the playlist of choice and then sync that to my phone where space actually does matter.

If you’re going full MP3 I would reconsider getting rid of the physical CDs, they may be fine for listening (depending who you ask of course) but it is not a good archival format.

Damn! i’m must admin i wasn’t able to tell the difference on any :frowning:

128 should be transparent to me, so I suspect you included one at an even lower rate.

At this point I’m keeping 128k rips (iTunes AAC on that computer, MP3 on Linux) + media for physical CDs, and whatever comes off Amazon MP3. Higher bitrate stuff gets recoded to 128k en route to portable devices.

Wait, you expect me to listen to John Cage - 4:11 in a less than perfect quality and not recognize the difference? I refuse to take part in such an abysmal experiment, sir, this is exactly why this internet will never take off.

I agree a better song should have been selected that had more range and less synthesized sounds. None-the-less, I could hear a difference on my setup. I have USB cable connected to a Nuforce Icon2 and the Windows 7 passes it as 24 bit / 96000 Hz. The speakers connected to the Icon2 DAC are Paradigm Mini Monitor v.5.

I ripped my collection to FLAC, under the assumption the equipment I use to listen to my music will likely increase in quality, so using a lossless format means I wouldn’t have to rip again to hear a better sound.

I couldnt get the files to play. Just loaded forever. I will admit that on most equipment I use I will not be able to hear a difference after the 160VBR (differs from track to track), but on my desktop I will. More importantly though, if you are getting rid of your CDs, please rip to FLAC. You can change that into anything and not lose anything. Also MP3 is pretty crap, it creates artefacts by itself.

One of the benefits of having noise-induced hearing loss: I can’t tell the difference between ANY of the files. Woot! They all sound just fine to me. But I still ripped all my CDs to variable 192kbps mp3s (before throwing them out) any way.

A poll with a value for the quality of each piece (instead of ranking them) would’ve been better. Now when people can’t hear the difference they still have to ‘rate’ them differently. This means your poll results are now not only determined by the perceived quality but also biased by the answering method. The best you can hope for is that people on average choose a random ranking for the tracks they can’t distinguish - something I doubt.

Have you ever converted an MP3 file to another format, say AAC? The result is significantly worse than encoding AAC from the original. My hearing is pretty average but CD -> MP3 -> AAC I ended up with malformed music. I think these conversions don’t stack well.

Today MP3 is something of a world standard. So this would only come into play if there are devices that require a different encoding.

Just adding my voice to those complaining about the material we’re supposed to rank. Without being given a reference track (by identifying the lossless one), I have no idea what the original synthesized drums should sound like. I can tell you which clip I like best but not which one is most faithful. An important distinction in my humble opinion. Now, nicely recorded acoustic drums/cymbals that have been artifacted via mp3 compression? Sadly, I do know what that sounds like.

Furthermore, to add to the complaints, the sample has also ripping errors. At the beginning of every version, in the pauses, you can clearly hear scratching introduced by missed sectors while grabbing. That is imho the worst offense of a lot of DL music. There’s no point in all the fancy lossless formats (flac, wv, ape) or high-quality lossy formats (AAC, OGG, MP3-320k) if the grab is ridden with audio gaps.

To me two of them sounded quite good, the other three were okay, but didn’t seem to have as much depth… but I couldn’t really point out why. I flipped back and forth between them for a bit, but gave my best guess as to which was good/bad :stuck_out_tongue:
I am not an audiophile, and I have terrible speakers, so it probably didn’t help much!

This test asks us to compare each track with what we expect to hear. How can that be meaningful with a pop recording, which has no real-world referent?

If you want to hear MP3 fail most obviously, listen to recordings of a capella chorus or string orchestra. At mid to high volumes (of the performers, not the playback device) you can often hear what used to be called “flanging”.

Also: you obviously never listen to opera, or other continuous music split (gaplessly) into tracks. Everyone can hear the glitch at each track change, it’s that obvious. This glitch is an artifact of the mp3 encoding format, and it cannot be corrected.

Also: since you’re including FLAC, including uncompressed (WAV) obfuscates the question.

Personally: Since I frequently edit and/or mix, I store as FLAC and downsample to suit the playback device. I can’t hear the difference between 192 and FLAC, but re-encode that 192 a couple of times and the result is disgusting.

I can’t hear the difference, until I can…
that is, on mediocre headphones with on-board audio, listening to over-produced 80s pop/rock like this, I cannot tell the difference between the encodings. Using a DAC to a decent high fidelity system and listening to opera, the lower quality encodings are blatantly horrible.
Like most others here, I keep archival copies for listening when concentrating on the music, and everyday copies for the mp3 player, car, etc. In the old days that meant LPs and cassettes, now it means FLAC and mp3… truth be told, I have mp3s made from cassettes made from LPs that no longer exist…

Your nephew should be punished for not keeping a lossless copy while ripping your CD collection :slight_smile:

I can’t say I generally hear a huge difference on headphones, but when I listen on real speakers (70s Sansuis) through a real amp (http://tubedepot.com/diy-k12g.html) at a decent volume, mp3s are lacking. This is especially true compared to vinyl. I personally didn’t really get why Black Sabbath were so heavy until I cranked it up on vinyl, at which point it was clear the mp3s and radio cuts were mere shadows of how it should really sound.

The point being is that I’d wonder if mp3 vs. lossless might be a bit too headphone focused.

They all sound exactly the same to me.

If you want a real torture test to compare encoding methods, while still being something fairly recognizable, go grab the Peter Gunn Theme off the Blues Brothers soundtrack. All that brass wreaks havoc on inferior encoders.