WAV MP3 Comparison

[dintymoore]

This is a song I did on my Roland SC-8850 SoundCanvas module. It's just the sequenced MIDI tracks, not finished, but I need to see what people think of different formats. For some reason when I record the SC-8850 to the stereo L/R track it sounds to me, way inferior to when the module is being triggered by the computer, so I'm not real happy with the sound. If I were to "finish" it, the cymbals, at least, would be replaced by real ones, etc...

Here it is as a mp3: http://musicmusicmusic.cn/audiotestmp3.html

Here it is as a 16 bit/44.1 WAV: http://musicmusicmusic.cn/audiotest.html

And here are both of them on one page: http://musicmusicmusic.cn/audiotestwavmp3.html

I'm wondering: Do you hear a difference in the WAV compared to the MP3?
Does it take long to load the WAV? Is it worth the larger memory for the WAV?

And on the page with both versions, did that work ok or was it a chaos where both started and it was crazy? Of course you can play both at one time.

I did this on a Mac OS X 10.4.11 so please let me know if you notice anything snakey on a Windows computer.

 

[cantthinkofname]

i thought the wav sounded richer but more tired than the mp3, and chaos when trying both versions at the same time, and using windows, am still trying to figure out your first question

aside from that very cool song

 

[mattr]

Well, you can learn nothing useful from any comments you receive whilst you are telling people which one is which. As soon as people know which one "should" be higher qualitiy, a placebo effect comes into play and they 'hear' differences because they expect (or think they should) be hearing them.

At the very least you should bounce the MP3 back to WAV and name them test_A.wav and test_B.wav (or similar). Better then would be to invite people to load them into their favourite ABX test software (foobar does the job!) as I'm about to do, and try and pass judgement there.

e.g.
No offense to cantthinkofname, but...

i thought the wav sounded richer

...is not even the kind of difference you would expect to hear when comparing MP3 compression. What is 'richness'? If its anything like the infamous and elusive 'warmth', it has a lot to do with the low-mids and mids, yet the effects of MP3 compression are usually most noticeable on the top end, and if anything were to happen to the mids with MP3 compression they are likely to lose some detail and muddy up a bit (which could actually make the MP3 sound 'richer' to the untrained ear).

Lossy compression rarely makes much tonal difference, apart from with really low bitrates (e.g. 96kbps and below) where a lot of the high end is lost. By the time you get to 320kbps, the only givaways could be minor compression artifacts, tiny losses of detail in things like reverb tails and sounds that are buried / masked in the mix. MP3 encoding is very clever - its algorithms target parts of the signal first which are deemed to be less audible to a listener (and so less noticeable of they are heavily compressed).


This raises the point that not all MP3 encoding is the same. Encoders sometimes give the option of fast encoding or high quality encoding, and the default is often the fast one. I've heard examples of two 128kbps MP3 files encoded from the same source... one was indistinguishable from the original in a quick listening test, whilst the other sounded more like 96kbps.


I'm a strong believer that 320kbps MP3 compression is fairly transparent in most situations. Heck, 192kbps is indistinguishable by many people.

On the matter of compressiong music for a website and streaming, I wrote a quite post a few months ago...

Despite all the hate that it receives, MP3 is a great format for music distribution. The compression can be more than transparent on well encoded files at respectable bitrates. I'd usually try to go for a minimum of 192kbps.

If these MP3s are going to be streamed with an embedded player then I wouldn't go any higher than 128kbps. You will hear some audible sound quality loss, but many people still do not have internet connections fast enough to stream anything greater than that without interruptions. Pleasing the largest audience possible should be your main priority, and so this may involve making some sacrifices to quality.

People complain about the quality of the myspace player, but it doesn't a good job right? The majority people with an 'ok' internet connection can listen to it and get a good idea of the music that an artist has to offer. Its only meant as a taster anyway.

You could then offer the chance to download the MP3 in a higher bitrate (192kbps or 256kbps), or develop the embedded player to allow users to chose the bitrate the music is streamed at depending on the speed of their internet connection.

So, in my opinion... Get LAME 3.98.1 and encode some good 192kbps MP3s for your website. Save the WAVs for a CD.


EDIT: I've just ABX'd the two files in foobar and honestly couldn't perceive a difference between them. I suggest you give it a go.

 

[Chibi Nappa]

I dig your 90's groove. Good song.
And mattr is correct. The only way to do this is to not tell people what they are listening to. I will say in the past I have identified 320 kbps Mp3s and anything lower is just a dead giveaway.

Does it take long to load the WAV? Is it worth the larger memory for the WAV?

I'm using Verizon Fios and the WAV plays just about as quickly as the MP3.

And on the page with both versions, did that work ok or was it a chaos where both started and it was crazy?

Firefox with Ubuntu Linux ran the page fine. No double-song chaos. All you need is a stop button so I don't have to reload the page to end song playback.

 

[cantthinkofname]

no offense taken mattr actually appreciate all that you've written because i'm learning from it so thanks

 

[dintymoore]

Thanks for the nice comments everybody. I am learning a lot from this.

I remember seeing Paul McCartney on Larry King, and Larry said something like "you're doing this for the people, right?".

And Paul's answer was "no, it's actually self indulgent, I do it for myself".

I thought it was a funny question for Larry, and suspect he knew what Paul's answer would be, especially since years ago I heard Larry say "I don't know anyone at the top of their field who isn't doing it for themselves".

So in some ways, even if no one but me can hear the difference between an mp3 or a WAV I want it to be top notch for me.

Might seem self centered, but that's what I want from an artist. Looking out at the audience and saying "Oh they want a cowboy dance, that's what I'll do" or "they can't hear the difference so I'll just bring my lighter amp to the gig" is bullshit. That's bad Las Vegas thinking.

Here's what I've noticed too: an mp3 and a WAV might sound the same to many, but for some reason I think poeple, without knowing it, get worn out by cheap formats, even though it might sound the same at first.

So why did I post this thread then?

Because I like to hear what other people think.

 

[mattr]

Fully understood dintymoore, but can I just ask one thing of you ... ?

And this is not me trying to promote selling lossy compressed, lower quality music, as that's silly - if you're selling it, offer it in as high quality as possible. But if you're distributing it on a website for free I propose that MP3 is a much more convenient and suitable choice.

You may believe you can hear a difference which I'm not totally disputing as some people can, however a lot of people can't and it would be interesting to actually see how your ears hold up in proper test conditions.

I was once absolutely convinced I could tell the difference between OGG Q5 and FLAC, but when I actually tested it out I couldn't perceive even the tiniest detail. Its very easy to hear differences because you expect them to be there. So, just try this one thing and I'd be very interested in hearing how you get on (please be honest!)...


Grab your favourite set of cans and find a Windows PC (I noticed you said you were running Mac - sorry but I don't know any ABX tools for OS X).

Download and install foobar2000...
http://www.foobar2000.org/

...and the ABX plugin, if it doesn't come with it...
http://www.foobar2000.org/components/view/foo_abx

Load the two files (MP3 and WAV) into the playlist, highlight both of them, scroll down to Utils, then select 'ABX two tracks'.


After clicking OK to where the temp files should be saved, you'll then be presented with a daunting looking test window. No longer do you have the comfort of knowing which is MP3 and which is WAV, they're now just named A and B. The challenge is to flick between A, B, X and Y until you make your mind up about how A and B match up with X and Y (then click the appropriate button).

You probably need to repeat this test 5 or 10 times (they're randomly swapped round every time) which then gives you a fairly clear readout of how accurately you could match them up and the probability that you were guessing each time.


Trust me, the results can be surprising.

The first time I tried it I thought my ears were deceiving me! There was some level of denial involved!

 

[dintymoore]

Good idea but I have no access to a Windows pc... I've spent less than 1 min on Windows and don't know anyone that has one... but it sounds like a great idea.

My concern is that I want to get the best sound out there possible. Every opportunity is an audition and even if I give something away for free on a website I'd like it to sound the best that it can.

I have a lot of trouble accepting any of the digital recording, it all sounds funny to me.

It may sound like I'm Cosmic Nature Boy, but when I go out in the back yard and hear the wind ripping through the trees, there's tons of treble, but it's feathery, it doesn't hurt at all. And thunder sounds way, way lower than I've heard any recording sound.

That's what I'm trying to get, or at least something close.

 

[cantthinkofname]

thanks to dinty & jeff i think i've made an important discovery!!! this made my whole day!!!!!! is it true we should pay close attention to live sound? & the 'arena' the live sound is in. could a small boxed room ever be the same as the big downstairs in a bar? etc, now i feel like i discovered the moon!

p.s. dinty i checked out the musicmusicmusic site out of curiousity, is that you?!? if so, your great, very great, if i could ever sound that good, that would be awesome

 

[Bristol Posse]

There's also the consideration of what people are listening on

Most computer desktop speakers are not going to give you a full 20hz - 20khz range, they'll cut off below about 80-100hz and above about 16k. So if the mp3 compression in higher bit rates is above the 16k mark it won't make any difference because that wouldn't even get played on the original WAV file anyway on such a sytem because the speakers aren't caple of reproducing those sounds. Lap tops and ear buds are even worse, generally speaking, so putting a horrible lossy format through horrible speakers that can't reproduce the very high and very low frequencies anyway, won't be very noticable until you get to bit rates on the mp3 below 96 kbit/s in my experience. I think that most radio stations cut everything above 16khz anyway and it's really not terribly noticable, except to maybe babies and dogs and bats who can maybe tell something is missing

I encode everything to 128kbits for mp3 when I'm posting on my own site to be considerate to slower internet connections but still keeping a reasonable playback quality, while assuming that most of the people will be listening on cheap garbage speakers hooked up to their computers anyway. But if putting on a CD or selling then WAV is the way to go.

 

[dintymoore]

Bristol I listened to that song and that's really good - I liked the arrangement, good tonal variation, and I liked it when the bass came in on verse 2 and introduced the subtonic. Good feel/vocals/song. Had a Natalie Merchant vibe to it, anyways that's what I thought.

You know when you make a mp3 at lower kbits it doesn't just chop off the "ends", the highs and lows - it screws up everything in between... a lot to my ears anyways. I notice it most in the hardest instrument to record - cymbals.

cantthinkofname - no matter if you're playing an acoustic instrument or electronic instrument, in the end, you're playing the room... you're carving that air. So ya, the room is super important. Ideally you have a room (like I don't) that has high ceilings, 10' min, and that makes the "tails" of notes sound feathery. In small rooms the sound stops abruptly and doesn't sound so good. That's where a good reverb helps because you can fake a big room in a small room.
Yes that's my website and thanks for the comments!

Here's what I really think about this mp3/WAV... all that:

We "hear" from 20Hz to 20KHz, but the main info is in a much narrower mid range - like an old telephone. But scientists in Russia have shown that we can sense up to 200KHz (zowie). Over half of the sound of a cymbal is above 100KHz!
I think there's an aura, electrical I guess, around us, and we "feel" music way, way above 20KHz through that aura. It has a massive effect on how we feel the music. Ever notice that you can be near a club and hear live cymbals and you know it's a live band? I've never, nor probably will be, fooled that a recorded cymbal is a live cymbal, because of all the missing info.
So that's why, even if people say they don't hear a difference in the lower version mp3's, I know that for people to feel the music it needs to be in a massively better version - WAV's and AIF's at 16 bit/44.1 are nowhere near what's needed - more like 256 bit/512.
I predict they will come out with stuff, and they'll have it at Sears, and it will sound like a violinist is standing right next to you. And people will say "remember when they used to say we only heard from 20 to 20?". And regular Joe's will buy it because they will be able to feel it.
You can't go by what most people say, because they're going by their ears and music isn't about sound, it's about feel. The sound is very, very secondary.

 

[Bristol Posse]

^^
Thanks for the props on the song I'm going to be re-reocrding the vocals and mastering the whole album in the next few weeks ready for my adoring public

We "hear" from 20Hz to 20KHz, but the main info is in a much narrower mid range - like an old telephone. But scientists in Russia have shown that we can sense up to 200KHz (zowie). Over half of the sound of a cymbal is above 100KHz!

That's why cymbals are so hard to get right (and why I use toontrack samples so all i have to worry about is triggering them, a pro had to deal with recording them). Because pretty much any mic, no matter how pricey, will only respond up to 20khz, maybe 22khz at the very outside and this response range hasn't changed in decades despite technological advances. And even something like KRK Exposes at $1500 per speaker will only respond and play back up to around 30khz. Even IF we may be able to feel sounds above 20khz and all the way up to 200khz, as those ruskies claim, nothing we have available is even close to being able to record or reproduce those frequencies. If we were able to get recording and playback gear able to even "hear" those frequencies, we'd also need a complete redesign of digital converters to run at 24/400 to get the nyquist math to "hear up to 200 khz. You'd also need about a TB of hard drive space per song to record a multitrack at that sample rate and an iPod the size of a server to store more than one song and play it back at those rates

Then again I now have more proecssing power in my phone than a roomfull of computers could manage in the forties so who knows

 

[SouthSIDE Glen]

But scientists in Russia have shown that we can sense up to 200KHz (zowie). Over half of the sound of a cymbal is above 100KHz!
I think there's an aura, electrical I guess, around us, and we "feel" music way, way above 20KHz through that aura. It has a massive effect on how we feel the music.

Aw, come on guys, we just spent a week or so clearing the forum of this nonsense in another thread just a couple of weeks ago, let's not bring this kind of hoodoo voodoo non-science back into it again.

Eve notice how it's always some obscure alleged "Russian scientists" that are behind this kind of stuff, right along with pshychokenetics and Tesla causing the Tunguska explosion and other such stuff, and never anybody from MIT or Bell Labs or JBL or any actual real audiologist or physicist conducting an actually scientifically sound experiment?

Here's a couple of points from actual, real physical science:

- There's a reason why when you look at a spectrograph of your average sound source - including a cymbal - that it always slopes downwards to the right. The higher the frequency, the greater the energy required to give it sufficient amplitude. Even *IF* a cymbal emitted sound waves at 100kHz, it would be at an extremely low, close to immeasurable level, no matter how hard you hit the cymbal. Over half the energy emitted by a cymbal is well below 10kHz, nowhere near 100kHz. Anything that *may* be left - if any at all - at 100k or above would be way too negligible to have a "massive effect" on anything in or surrounding the human body, Russian aura or not. The starting energy just is not there.

- There is also a direct scientific relationship between frequency and propagation loss in any medium, including air. In average humidity air, this translates to about -1800dB of attenuation at one kilometer. This means that even a 180dB sound - enough to make your ears bleed and cause deafness - at 100kHz would only make it less than 40 feet before dying to complete silence. Even if there were 100kHz information at a full 30dB of whisper - which is practically impossible given the energy distribution - it would only make it some 6 feet off the cymbal before dying out to nothing. And that's not even counting having to transfer through a wall or window, as it would when talking about being "near a club", even though such solid surfaces are going to absorb or reflect so much HF back into the room as to be virtually opaque at those frequencies. At the fraction of a dB or so that maybe *might* come off a cymbal, maybe it might last for a full millimeter on a good day.

G.

 

[Bristol Posse]

Eve notice how it's always some obscure alleged "Russian scientists" that are behind this kind of stuff, right along with pshychokenetics and Tesla causing the Tunguska explosion and other such stuff, and never anybody from MIT or Bell Labs or JBL or any actual real audiologist or physicist conducting an actually scientifically sound experiment?

I've also heard it from a bloke at the pub, who's brother knows this bloke, who's wifes best friends cousin is dating the guy who did this experiment

 

[SouthSIDE Glen]

I've also heard it from a bloke at the pub, who's brother knows this bloke, who's wifes best friends cousin is dating the guy who did this experiment

Wow, that's incredible! Would you believe I heard it from that bloke's barber? Man, talk about a small world!



I then told it to Kevin Bacon...

G.

 

[Bristol Posse]

Wow, that's incredible! Would you believe I heard it from that bloke's barber? Man, talk about a small world!



I then told it to Kevin Bacon...

G.

Yeah six degrees to bacon. I always suspected he was involved in these experiments. He's a lot more sinister than it would first appear

 

[cantthinkofname]

SouthSIDEGlen you obviously don't have to but you should really learn more about Tesla, being into what your into i'd be surprised you wouldn't admire and hang a pic of him up in your studio after learning more, i've been planning on hanging a huge one of him up in my kitchen pretty soon. And he's not from Russia, he's from Serbia and immigrated to America

 

[SouthSIDE Glen]

SouthSIDEGlen you obviously don't have to but you should really learn more about Tesla, being into what your into i'd be surprised you wouldn't admire and hang a pic of him up in your studio after learning more, i've been planning on hanging a huge one of him up in my kitchen pretty soon. And he's not from Russia, he's from Serbia and immigrated to America

I won't claim to be Tesla's biographer, but I have read plenty about him in the past, both fact and fiction. He was a cool guy with a lot of cool ideas and did some pretty cool things, no doubt. But I can't stomach yet another baloney conspiracy theory religion and Internet myth like the one surrounding him and espoused as "truth" in the majority of the books written about him these days.

I never said he was from Russia. I was referring to that same gang of wacko and possibly imaginary "Russian scientists" who claim the famous Tunguska explosion in Russia was the result of a high-voltage directed energy "weapon" that Tesla was supposedly working on here in North America.

G.

 

[dodgeaspen]

I know this will not help but I can't hear a difference on the speakers I'm using. But I will say it sounds good with 1 or both going at the same time.

 

[cantthinkofname]

cantthinkofname - no matter if you're playing an acoustic instrument or electronic instrument, in the end, you're playing the room... you're carving that air. So ya, the room is super important. Ideally you have a room (like I don't) that has high ceilings, 10' min, and that makes the "tails" of notes sound feathery. In small rooms the sound stops abruptly and doesn't sound so good. That's where a good reverb helps because you can fake a big room in a small room.
Yes that's my website and thanks for the comments!

Here's what I really think about this mp3/WAV... all that:

We "hear" from 20Hz to 20KHz, but the main info is in a much narrower mid range - like an old telephone. But scientists in Russia have shown that we can sense up to 200KHz (zowie). Over half of the sound of a cymbal is above 100KHz!
I think there's an aura, electrical I guess, around us, and we "feel" music way, way above 20KHz through that aura. It has a massive effect on how we feel the music. Ever notice that you can be near a club and hear live cymbals and you know it's a live band? I've never, nor probably will be, fooled that a recorded cymbal is a live cymbal, because of all the missing info.
So that's why, even if people say they don't hear a difference in the lower version mp3's, I know that for people to feel the music it needs to be in a massively better version - WAV's and AIF's at 16 bit/44.1 are nowhere near what's needed - more like 256 bit/512.
I predict they will come out with stuff, and they'll have it at Sears, and it will sound like a violinist is standing right next to you. And people will say "remember when they used to say we only heard from 20 to 20?". And regular Joe's will buy it because they will be able to feel it.
You can't go by what most people say, because they're going by their ears and music isn't about sound, it's about feel. The sound is very, very secondary.

thanks, i agree, & as for the room last night i was reading something & it said the same thing as you about when not having a choice using reverb to fake the room is what we should do.

i have been considering a small room with wooden floors in my house to eventually record in but now am thinking it might be smarter to use my garage because it is much larger with high ceilings, that along with added reverb might help & be preferable over the small room? Along with finding creative ways to get a perfected 'live' sound rather than following standard mp3?


SouthSIDEGlen, tesla is MUCH more than just someone who had a few cool ideas, i don't mean this offensively, but when i read what you wrote i about fell off my chair. your other comments tend to make sense but think there is alot missing, you are thinking more as a metaphysicist, not as an empirical scientist, not that that is entirely wrong either, but leaves more room for speculation and falling down the rabbit hole.

well thanks, this entire thread so far has been alot of help