MP3 versus WAV, FLAC, AIFF, or...

[KLH]

Many beginners ask questions like "which is best, MP3 or ?" I want to give my opinion and open this up to discussion. However, to answer this question properly, you have to understand basics about digital audio and then what works with most DJ software.

[I]Digital Audio Basics

Here's a quick primer on digital audio from the Wiki (http://en.wikipedia.org/wiki/Digital_audio):

A digital audio system starts with an ADC that converts an analog signal to a digital signal.The ADC runs at a sampling rate and converts at a known bit resolution. For example, CD audio has a sampling rate of 44.1 kHz (44,100 samples per second) and 16-bit resolution for each channel... The digital audio signal may be stored or transmitted. Digital audio storage can be on a CD, a digital audio player, a hard drive, USB flash drive, CompactFlash, or any other digital data storage device... Audio data compression techniques — such as MP3, Advanced Audio Coding, Ogg Vorbis, or FLAC — are commonly employed to reduce the file size... The last step is for digital audio to be converted back to an analog signal with a DAC.

So there you have it, digital audio is a representation of analog audio that is limited by how the audio is represented. In the usual form of Pulse Code Modulation (aka PCM), audio is represented by a sample with fixed resolution, called bit depth (16bits for CDs), and a rate of samples, called the sample rate. Together, the sample rate and bit depth determine how accurately the original sound is captured as digital audio. At this sample rate and bit depth, stereo sound uses about 605MB per hour. (http://24bit.turtleside.com/pcm.wav.file.sizes.pdf)

There are many files formats that contain native PCM audio - WAV (popularized on Windows) and AIFF (popularized on MacOS) are just two popular file formats. The common characteristic of PCM audio is that it is large in size. Back in the early 80s, CDs were created to store and playback PCM digital audio.

In the mid 90s, data compression (http://en.wikipedia.org/wiki/Audio_compression_(data)#Audio) techniques became popular that to reduce the amount of data being stored in files. Keeping this simple, two major ways emerged. One was to reduce the amount of data and still ensure that ALL of the original data was recreated - called lossless. The other way was to reduce the data, but not mind so much if some of the original data was lost (reasoning that you wouldn't miss it anyway) - called lossy. Lossless would reduce the file sizes up to 50% (302 MB per hour) and sound the SAME as the original recording. Lossy methods could reduce the file sizes by 80%+... but would reduce sound quality at high compression rates.

Through time the two coding/decoding techniques (called codecs) that became very popular were FLAC for lossless and MPeg 1 layer 3 (aka MP3) for lossy. There are many other codecs (http://en.wikipedia.org/wiki/Audio_file_format) for both lossless (i.e. Apple Lossless, WMA Lossless, etc.) and lossy (WMA, M4P, Ogg, etc.).

Skipping how it came to be, let's just cut to the chase and state that MP3 is the most popular of digital audio file formats.

DJ Software Context

Music exists as digital audio files in a variety of codecs - WAV, AIFF, FLAC, MP3, AAC, and more. Many DJ applications playback MP3, WAV, and FLAC. Others playback MP3, AIFF, and AAC. Others still playback MP3, FLAC, and AAC. See the pattern? MP3 is ALWAYS played back by DJ apps. For that reason, most DJs tend to use MP3 as their format of choice.

So Which Is Best?

In terms of recreating the original analag audio with minimal data, technically lossless is best. It has the accuracy of PCM when decompressed AND reduced data size when stored. Still, all DJ apps use MP3s as an accepted format.

So what most DJs tend to do is use the highest data rate in the MP3 format - 320kbps Constant Bit Rate MP3. Doing so is the best compromise between file size and audio quality (assuming that the original recording is of high quality).

So now you know... and knowing is half the battle.

http://cache1.bigcartel.com/product_images/74312121/knowing.jpg

-KLH

[Atomisk]

Good guide. Sticky?

flac/wav ftw. Everything should be in lossless. I see no reason why not - HD space is insanely cheap nowadays.

[miX_]

I buy and rip all my CDs in WAV. External HD [backups] are cheap stuff, good to have. I like to think I'm supporting better quality tracks (from a pure mixdown / mastering standpoint) if labels are noticing people are buying in higher quality. That's what I'd like to promote.

[snoborder101]

The thing is, where are you guys buying your music from? For the longest time I've been buying from itunes 256 AAC. I'd like wav files, but it's expensive enough stocking a library full of music. Beatport is already more expensive than itunes, and then they charge you even more for wav files.

[mrkleen]

Unless you are playing on function one level systems...you are NOT going to hear the difference between a well encoded 320 mp3 and a wav.

[Atomisk]

Unless you are playing on function one level systems...you are NOT going to hear the difference between a well encoded 320 mp3 and a wav.

Lolwut I can hear the difference between 320 and lossless in my HDJ-2000s and those are NOT accurate headphones.

[DJNR]

Good article! I'm waiting for Mostapha's appearance and commentary on this issue :P

[NPC]

Lolwut I can hear the difference between 320 and lossless in my HDJ-2000s and those are NOT accurate headphones.

No, you just think you can. Really mp3@320 rolls off well above most human hearing. Most people can't hear above 17 or 18KHz.

I have a test you can use, I'll dig for it or maybe just make my own. One sec.

[Hamza21]

Lolwut I can hear the difference between 320 and lossless in my HDJ-2000s and those are NOT accurate headphones.

What are saying? You can hear a difference between 320 and lossless version of the same track? Anybody can do that. Or are you saying you can hear a difference an 320 track of one song and lossless track of another song? I highly doubt that! In order for that to be possible you need some frame of reference. If you never heard a song before you definitely can't tell between two different tracks. I've heard Award tour by ATCQ 1,000 of times I couldn't tell the difference between an 320 version or wav version unless played one after the other. I highly doubt you can.

[mostapha]

Good article! I'm waiting for Mostapha's appearance and commentary on this issue :P

Happy to oblige.


No, you just think you can [hear the difference]. Really mp3@320 rolls off well above most human hearing. Most people can't hear above 17 or 18KHz.

I have a test you can use, I'll dig for it or maybe just make my own. One sec.

Last time I did a blind test, I correctly identified the lossless file 100% of the time for 5 tracks with a maximum of 3 listens to the first 5 seconds of each version. It wasn't purely scientific, but the worst score I've gotten was 80%, and that was only allowing 2 listens each to tracks I'd never heard before.

Here's the thing…hard drives, flash storage, and CD-Rs are cheap. The question isn't even whether there's a perceptible difference. There is a difference, hard drives are cheap, and mp3s won't be around forever…so why bother with mp3s? They have no advantages except being supported by everything. All but the cheapest garbage supports wav or aiff as well.


The thing is, where are you guys buying your music from? For the longest time I've been buying from itunes 256 AAC. I'd like wav files, but it's expensive enough stocking a library full of music. Beatport is already more expensive than itunes, and then they charge you even more for wav files.

Boo f***ing hoo. Go look at vinyl prices and come back to me. Beatport is probably the cheapest of the places I shop because most of the rest are either smaller or in the UK.


In order for that to be possible you need some frame of reference.

For me to be 80-100% accurate, yes. But I'm better than random listening to other people's iPods in my car (ripped to wav and put on the iPod vs. bought at iTunes). Never really kept track of it, so IDK if it's significant.

[Era 7]

Good guide. Sticky?

flac/wav ftw. Everything should be in lossless. I see no reason why not - HD space is insanely cheap nowadays.

pff :wang: aiff is where it is at. lossless quality + tags, is recognized by almost any recent software AND is carried as a standard at beatport. WIN. :slayer:

but honestly any lossless format will do. my philosphy is if it is available in lossless then why not? HDDs are huge these days. only a matter of money.

[mrkleen]

Lolwut I can hear the difference between 320 and lossless in my HDJ-2000s and those are NOT accurate headphones.

We are not talking about headphones in your bedroom. We are talking about playing those songs out on loud sound systems in clubs.

If I can get lossless, I do. But 99% of the promos I receive are 320 mp3s. I am on many of the same promo lists as top techno DJs, so if 320 is good enough for Chris Liebing and Carl Cox, it is good enough for me.

[mostapha]

Yeah…it's better to play a bootleg of an awesome track that you recorded onto a walkman in the artist's basement than to not play it…but that's not what this is about either. If it's available lossless, mp3s provide zero benefits. Those tracks aren't available lossless (to you, in that way), so the comparison is moot.

[princesultan]

yup. played on one the other day, 320 still sounds great and all the big time djs play them. no one can hear the difference except maybe tony andrews himself, lol.

[Finnish_Fox]

We are not talking about headphones in your bedroom. We are talking about playing those songs out on loud sound systems in clubs.

So you are acknowledging he can hear the difference in his headphones?

Well, there goes your point... :P

[Finnish_Fox]

yup. played on one the other day, 320 still sounds great and all the big time djs play them. no one can hear the difference except maybe tony andrews himself, lol.

...or anyone that just listens closely.

[mrkleen]

Yeah…it's better to play a bootleg of an awesome track that you recorded onto a walkman in the artist's basement than to not play it…but that's not what this is about either. If it's available lossless, mp3s provide zero benefits. Those tracks aren't available lossless (to you, in that way), so the comparison is moot.

The benefit is it doesnt take a hour to download a track you received from a mate in the UK.

I am all for the theoretical world where Tony Andrews lives - where everyone has 50K dollars for a pair of speakers and flawless tube amps and processing and wav files are the bare minimum, but in the real world, 320 is what most people are playing and it does just fine for 99% of all applications.

[tekno_violet]

Snoborder, on beatport just stockpile the music you want in your hold bin then after a few weeks find a promocode online for around 25% off and buy up whats in your hold bin. That'll make up for the extra cost for loss-less.

[mrkleen]

So you are acknowledging he can hear the difference in his headphones?

Well, there goes your point... :P

Unless you have 500 people inside his headphones, NO that doesnt refute my point at all. :rolleyes:

[mostapha]

The benefit is it doesnt take a hour to download a track you received from a mate in the UK.

Weird. It doesn't take that long to download an 80MB file for me. Maybe if someone's internet sucks. Still…how much of a hurry are you in? Still quicker than mailing them.

[Finnish_Fox]

Unless you have 500 people inside his headphones, NO that doesnt refute my point at all. :rolleyes:

What does 500 people have to do with him being able to distinguish between a .WAV and 320kbps? Wasn't the issue whether a difference could be heard between a lossless .WAV and a 320kbps, with you arguing that no difference can be heard? Seemed to me that, later, you acknowledged a difference could be heard but not on a huge system. :shrug:

Either way, 98% of the general public won't notice a difference.

[mrkleen]

What does 500 people have to do with him being able to distinguish between a .WAV and 320kbps? Wasn't the issue whether a difference could be heard between a lossless .WAV and a 320kbps, with you arguing that no difference can be heard? Seemed to me that, later, you acknowledged a difference could be heard but not on a huge system. :shrug:

Nope. Maybe you should go back and read my first post


Unless you are playing on function one level systems...you are NOT going to hear the difference between a well encoded 320 mp3 and a wav.


Either way, 98% of the general public won't notice a difference.

So now you are agreeing with me? Nice one. :tup:

[Finnish_Fox]

Nope. Maybe you should go back and read my first post

Was this your first post? And here was his response to the post I've quoted:



Unless you are playing on function one level systems...you are NOT going to hear the difference between a well encoded 320 mp3 and a wav.


Lolwut I can hear the difference between 320 and lossless in my HDJ-2000s and those are NOT accurate headphones.

Unless I am missing something...

You: "You cannot hear the difference between a .WAV and a good 320kbps .MP3 unless it is on this one particular soundsystem."
Him: "Umm, I can hear the difference in my headphones, actually."
You: "Yah, but thats not a club with 500 people in it."

So, while your point may have been that the overwhelming majority of people will not notice the difference on a large/powerful/good soundsystem, its not really what you said, at least from what I can tell.

Am I missing something? :shrug:


So now you are agreeing with me? Nice one. :tup:

To play semantics, 98% of the public means that some people can tell the difference which isn't exactly what you said in the first place.

Sounds like we are on the same page overall.

[mostapha]

Either way, 98% of the general public won't notice a difference. 98% of the general public won't notice the difference between a good track and an awesome one either……and it's their loss. Quality should be no different.


You: "You cannot hear the difference between a .WAV and a good 320kbps .MP3 unless it is on this one particular soundsystem."
Him: "Umm, I can hear the difference in my headphones, actually."
You: "Yah, but thats not a club with 500 people in it."

So, while your point may have been that the overwhelming majority of people will not notice the difference on a large/powerful/good soundsystem, its not really what you said, at least from what I can tell.

Am I missing something?

Yes. MrKleen used "you" as a generic 3rd person pronoun–which should technically have been "one"–because American English has pretty much dropped that pronoun from its vocabulary. You inferred the 2nd person singular pronoun.

I also disagree with his post in that it doesn't take nearly a Function One system to hear the difference. I can hear it on my HD-25s and KRK RP5s. I can also hear it on NS-10s and nice, big main monitoring systems. I think I'd be less likely to hear it on Function One, but only because the only ones I've heard are so loud I wouldn't listen without earplugs. And the earplugs obscure the system's noise floor……which would make hearing reverb tails disappear more difficult. I don't know if I could ABX the difference while wearing plugs. And I can't do it at club volumes because I'm not willing to listen to anything that loud without protection.

Then again, maybe I'm missing something. At any rate…it's not relevant.

We've all come to basically the same agreement (that there is a difference and that it's audible to a non-zero amount of people) and just differ on what compromises we're willing to make in the hard drive space and transfer time vs. objective quality battle. I know what I want to do, and I don't really care what other people do except that I haven't yet heard a really good argument for lossy compression. Maybe 10 years ago…but not today.

Lets just say that if Apple made their entire iTunes library available as DRM-Free lossless files…they'd make a lot of money, because I'd probably re-buy my entire listening collection of hip hop and country in one fell swoop.

(disclaimer: I don't just listen to hip hop and country when I'm not DJing, but it's funnier that way)

[Finnish_Fox]

98% of the general public won't notice the difference between a good track and an awesome one either……and it's their loss. Quality should be no different.

True that.


Yes. MrKleen used "you" as a generic 3rd person pronoun–which should technically have been "one"–because American English has pretty much dropped that pronoun from its vocabulary. You inferred the 2nd person singular pronoun.

Well, the other guy did. I was just supporting what he said... Even still, you still doesn't make it 100% of the time. Haha.


Then again, maybe I'm missing something. At any point…it's not relevant. We've all come to basically the same agreement (that there is a difference and that it's audible to a non-zero amount of people) and just differ on what compromises we're willing to make in the hard drive space and transfer time vs. objective quality battle.

/Thread.

[mrkleen]

Unless I am missing something...

You: "You cannot hear the difference between a .WAV and a good 320kbps .MP3 unless it is on this one particular soundsystem."
Him: "Umm, I can hear the difference in my headphones, actually."
You: "Yah, but thats not a club with 500 people in it."

So, while your point may have been that the overwhelming majority of people will not notice the difference on a large/powerful/good soundsystem, its not really what you said, at least from what I can tell.

Yes you are totally missing the point. My point was that unless you are listening on a very very good sound system, or on a critical listening device like a pair of headphones an inch from your ear drum - you will not be able to hear the difference...and even if you can, the difference will be minuscule.

Maybe you play in rooms full of audiophiles, on pristine sound systems....but I doubt it. If anyone on this board has experience with this, it is Prince Sultan - who regularly plays at Stereo in Montreal, one of the 5 or 10 best sound systems in the world.

If HE can not hear the difference and (correctly) states that most guest DJs who he opens for are playing 320 mp3s - then I guess I dont get your point.

Take your average clubber, plop him down in the middle of a dancefloor at 5 a.m. - give him 100 db of sound ratting through a mediocre sound system - 5 Vodka Red Bulls and 2 pills - 1 girl whispering in his ear and 5 mates yelling at him....then add 500 other punters doing the same. You would NEVER hear a difference...and neither would any audiophile.

A bunch of bedroom DJs on this board claiming they can hear a difference between the wav and mp3 version of a song they are listening to in their headphones means NOTHING in the real world.

[Finnish_Fox]

Yes you are totally missing the point. My point was that unless you are listening on a very very good sound system, or on a critical listening device like a pair of headphones an inch from your ear drum - you will not be able to hear the difference...and even if you can, the difference will be minuscule.

That definitely wasn't clear the way you wrote it... even Mostapha misread it.


Maybe you play in rooms full of audiophiles, on pristine sound systems....but I doubt it. If anyone on this board has experience with this, it is Prince Sultan - who regularly plays at Stereo in Montreal, one of the 5 or 10 best sound systems in the world.

I do... First and foremost, I am my own primary audience. If I am not making mixes, I enjoy, I'd stop mixing. I mostly listen to my own mixes in the car or on personal listening devices, so I notice the difference, personally.

Would that stop me from playing 320kpbs files out and about? No... precisely because I know that 98% won't notice it and most soundsystems are not good enough to hear it.


If HE can not hear the difference and (correctly) states that most guest DJs who he opens for are playing 320 mp3s - then I guess I dont get your point.

So, just because one person who regularly plays on a superb sounndsystem can't tell the difference, no one can? Not sure I follow that logic. My point is that the difference is audible to some people in certain situations. You didn't exactly specify the situation in which the difference is noticeable or not which is where the disagreement arose.

Most people won't notice it on most systems...


Take your average clubber, plop him down in the middle of a dancefloor at 5 a.m. - give him 100 db of sound ratting through a mediocre sound system - 5 Vodka Red Bulls and 2 pills - 1 girl whispering in his ear and 5 mates yelling at him....then add 500 other punters doing the same. You would NEVER hear a difference...and neither would any audiophile.

http://t0.gstatic.com/images?q=tbn:ANd9GcSzCiPa14SS-9eUUPi-IShR4yRBa0Qdk51WEHbBFZL_gLjizfc5&t=1

You mean on someone would have a tough time distinguishing track quality through a mediocre source (with or without booze and drugs)? Go figure.


A bunch of bedroom DJs on this board claiming they can hear a difference between the wav and mp3 version of a song they are listening to in their headphones means NOTHING in the real world.

It means nothing except that they notice a difference in their headphones which is precisely what the guy said.

...and get off your "I'm a big club DJ" high horse buddy.

[mrkleen]

...and get off your "I'm a big club DJ" high horse buddy.

....and get off your, I give "theoretical" advice to people because I dont have any practical, real world advice to give horse buddy.

The last thing kids out here need is a bunch of BS about worrying that the songs they get on Itunes and Beatport are not high quality enough to play out with. You guys keep worrying about bit rate...while the rest of us will go out make a living with our mp3s.

[mostapha]

And I've said more than once that a good enough song means "screw quality." But there's still no reason to choose worse quality if you don't have to.

You're right, though…it's not a big enough issue to fight about.

[Moss]

My personal opinion on this is that to really hear the difference in quality , you have to be in the perfect situation. If a club has a djm 800 for example & a decent sound system, you wont hear a noticeable the difference. Your sound quality will be as good as your least good piece of equipment (whether it's cables , sound card , cdj's , mixer , speakers ...). I wish I could get all the tracks I want on wav or aiff , but it ain't the case.

At a club event I did with my buddy , we made a test. He was using his sl-3 and I was using my sl-1. He was always clipping the output of serato and the mixer's channel. Even though his sl-3 is superior in sound quality then my sl-1, I still had a much cleaner output.

(Sorry if my english is not perfect , it's not my fluent language :))

[DJNR]

If you can house the better quality tracks (in regards to space and budget), great, play them because it doesn't hurt. If you can't, no big deal as long as you can get a decent version of a track.

[Smallz]

One thing I want to make note is that WAVs are a little annoying when it comes to tagging the track, so MP3, FLAC and AIFF are better in that instance.

That is all.

[princesultan]

it really is the same debate over and over and over again. there will always be those who claim that it must be .wav or nothing since the difference between 320 is "noticeably" different. if you're gigging, bottom line is this... if 320 is good enough for 98% of the touring djs out there, then it's good enough for anyone on this board. if you're just playing at home, then stick with what makes you happy.

[Nacho1000]

Hello. This is my first post in this fórum and I do hope it won't be the last.

There are some opinions in the thread in the sense that lossless format are a better choice cause storage is cheap nowadays. I think the same. But I wonder if cheap equipment can reproduce with minimal latency those heavy formats.

Any opinions on this?

[KLH]

Welcome to DJF Nacho1000! Great question.

Back in the early 2000s, CPUs needed to be strong in order to handle the heavier lossless codecs. Nowadays, CPUs are plenty powerful to handle any audio codec.

[DJ Elevate]

I haven't read all the pages of stuff, I'll just speak from experience. I've been using AIFF since I started getting serious and have had zero issue across multiple platforms (Traktor and CDJ standalone players and controllers). I did some research before I made this decision and I believe the format also has the best metatag support - which I use extensively to notate the "feel/vibe" of the track.

That said - it is a resource hog but this has been a gift and a curse. Curse in the sense that I'm always having to free up hard drive space because I primarily play what's on my laptop - I go to my external when I have to dig for a request. The gift is, there's something to be said about quality over quantity. I'm forced to constantly review my library and get rid of tracks that are old and/or I'm just not digging them anymore, so my primary playing library just becomes a better and better big curated playlist.

[Windows 95]

That said - it is a resource hog but this has been a gift and a curse. Curse in the sense that I'm always having to free up hard drive space because I primarily play what's on my laptop - I go to my external when I have to dig for a request. The gift is, there's something to be said about quality over quantity. I'm forced to constantly review my library and get rid of tracks that are old and/or I'm just not digging them anymore, so my primary playing library just becomes a better and better big curated playlist.How old is your laptop that you can't hold a few hundred gigabytes of music on it?

[light-o-matic]

Re lossless formats:

AIFF and WAV are both lossless, full size formats.. they take up around the same disk space for the same audio, but aiff is taggable, wav is not. Therefore if your software can read aiff, it's always the better choice.

FLAC is also lossless.. and taggable, and takes up around 30% less disk space.. so it's superior to WAV and AIFF. But only if your DJ software supports it.. some do some don't. The newest Pioneer decks and newest Rekordbox supports it (the last year or so), the older ones don't. Eg. CDJ-900 does not, XDJ-1000mk1 does not.. XDJ-1000mk2 does.

Not all lossless files are the same quality.. since the bit depth can vary (used to be always 16 bit.. which was the CD standard.. now sometimes you will get 24 bit releases) as well as the sample rate (used to always be 44.1K.. which is CD standard, now you will sometimes get 48K or 96K).. also sometimes DSD rips. These higher quality lossless files tend to be huge in size, where a CD quality (16/44) WAV or AIFF will be approx 10 megabytes per minute of music, a 24/96 version will be over 26 megabytes per minute. MOST files you will get are still going to be 16/44 or 16/48..

[Windows 95]

I highly doubt anyone can really hear the difference between CD quality & above CD quality.
96K needs to stay in people's DAWs.

[light-o-matic]

I highly doubt anyone can really hear the difference between CD quality & above CD quality.
96K needs to stay in people's DAWs.

I was very surprised to find that I could hear some differences between CD and 24/96 of the same material, but yea, basically I agree, it makes no difference in DJing. I do think that 48K vs 44.1K is worth it.. now that CD is losing relevance.

[DJ Elevate]

How old is your laptop that you can't hold a few hundred gigabytes of music on it?

13" Macbook Pro on the cheaper end. 128GB HD and almost half is taken up with host OS + apps.

[SlayForMoney]

I do think that 48K vs 44.1K is worth it.. now that CD is losing relevance.

https://en.wikipedia.org/wiki/Hearing_range

"he human range is commonly given as 20 to 20,000 Hz,"

Whoever started encoding their MP3 files in 48K needs to be brought outside and shot.

[mitchiemasha]

44 v 48! when it comes to CD losing relevance, going 24bit over 16 is a much better advantage than jumping to 48.

[Hanginon]

For those who would like to continue pursuing this - check out Hydrogenaudio, a group dedicated to the scientific comparison and advancement of audio codecs -
https://hydrogenaud.io/

Know this, the audio artifacts created when compressing a file to 320 CBR mp3 are far less than the phase modulation distortion created by a 15" woofer, when reproducing the mid range of a song, in a typical 15" two-way loudspeaker - so sing to the right choir.

[Alex Murphy]

I'm currently buying a few CDs because the mp3 versions I listen to don't cut the mustard. I am also taking into account that my next mix will partly be released as an mp3. As in making an mp3 of an mp3 is like doing a photocopy of a photocopy of an original. It just won't sound as good. I also scored a few CDs with extra tracks for the price of a single mp3.

As a side note, I wouldn't compare a digital artefact with that of a physical speaker. In the higher range, mp3 just don't sizzle like a tweeter would distort, and in the bass department, it just doesn't hit as deep or as hard... Finally there are some bad speakers, and there are some bad 320s.

[Ultra]

For those who would like to continue pursuing this - check out Hydrogenaudio, a group dedicated to the scientific comparison and advancement of audio codecs -
https://hydrogenaud.io/

Know this, the audio artifacts created when compressing a file to 320 CBR mp3 are far less than the phase modulation distortion created by a 15" woofer, when reproducing the mid range of a song, in a typical 15" two-way loudspeaker - so sing to the right choir.

^^Thats good to know!

[mitchiemasha]

As a side note, I wouldn't compare a digital artefact with that of a physical speaker.
I would. Comparisons are the easiest ways to explain things. I've often described it similar but used the room you're in as the worsening contributor. I know how bad a 15" driver is in a mid/top speaker, i'd not thought of using that comparison before, i like it.


and in the bass department, it just doesn't hit as deep or as hard
This is simply not true and there's no reason for it. A speaker simply moves moves backwards and forwards... as long as the MP3 is well produced, that driver will perform the same. There's enough sample points to which the redrawn curve could potentially null if inverted. Remember, it's all curves once it's converted back, it's the stepped drawing that confuses people, it doesn't exist.

[mitchiemasha]

This explains the steps issue

https://youtu.be/cIQ9IXSUzuM?t=5m49s

[light-o-matic]

44 v 48! when it comes to CD losing relevance, going 24bit over 16 is a much better advantage than jumping to 48.


This explains the steps issue

https://youtu.be/cIQ9IXSUzuM?t=5m49s


Great videos, he's a good explainer.. but he glosses over a very important aspect. He says, "there's only one band limited signal that passes exactly through each sample point".. and then goes on to say that any signal that does not pass through the sample points is "at or beyond Nyquist". What he's talking about is the need for a low pass filter.. both on the input signal when encoding from analog, and the output signal when decoding to analog (or, also when resampling).. that completely cuts off any signal above the Nyquist frequency.. which is HALF the sample rate. So, if your sample rate is 44.1K/s you need a low pass that thoroughly eliminates any signal above 22.05KHz. One of the major problems with the sound of early CD audio was the difficulty of designing a filter that was so sharp that it sufficiently suppressed signals at or above 22.05K, without having a negative effect on signals below 20K. Any effect below 20K obviously affects audible signals, any signal that leaks through above 22.05K results in artifacts (aliasing in the audible range). This is why the pro audio standard sample rate was 48K/s, not 44.1.. it gives a little more room for that filter to be effective.

Granted, now you have a lot more advanced converter designs, and the filters are a lot better than they were (depending on the gear you are using of course), so things are better. But that is the problem we are dealing with at CD sample rates vs pro audio rates.

As far as 16 bit depth vs 24, actually, when you are dealing with DJing, car audio and actually a lot of home listening and ****** listening too.. yea, of course 24 bit gives you a wider dynamic range and thus higher S/N ratio. CD/16bit audio has a theoretical dynamic range of 96dB, vs 144dB for 24 bit.. so BETTER, right? Well, a nice quiet room at home, or a quiet library.. is around 40+ dB.. a typical pretty loud danceclub plays music around 100-105dBA / 120-125dBC.. or so. So, a dynamic range of 85dB, TOPS.. in a library. But in a dance club, with people shouting, jumping around, stomping, clinking glasses, etc.. the ice machines, fridges, air conditioning, etc.. So probably you are looking at somewhere in the mid 60's, at best, so you have an S/N of 60 in the club. So you see, the diff between 96dB S/N and 144dB S/N does not matter when you're in an environment with an S/N of 60dB. Then the fact remains that the producer and mastering engineer isn't going to actually put too much detail down at -100+dB, even on the 24 bit version where they can do it.. because they know that nobody is going to hear it unless they're in a serious listening environment or good closed back headphones...

BTW the S/N of a really well pressed vinyl is only around 65-70, and that's when it's brand new.

[mitchiemasha]

Don't forget that 24 bit can get closer to clipping with out clipping. Very important in the DAW world. Adds a lot of headroom. It's not just about the details down at -100db. The precision is across the full dynamic range.

As for his explanation on the 1 band limited signal and when he mention "Nyquist" on the sine wave, it's more to do with everything been made up of sine waves and is shown better later in the video with the square wave. For the signal to move through a different point it would require the mathematically related higher frequencies, harmonics to be there (or not be there) to result in it taking that different path. As they aren't there (or are there) the resulting frequency will always take the same path. Any harmonic above Nyquist (for human hearing) would result in a change so minute that it makes no difference, at least not to us. A sine wave itself has no harmonics, it's a pure tone.

[light-o-matic]

It'll take the same path IF the filter cuts off completely at or below the Nyquist frequency. If it doesn't.. it won't. You are ignoring this important point. The only signal possible at 20Khz is a sine wave.. yes, because any other shape would by definition contain components above 20K. Hence the need for an absolutely strict filter of frequencies above 20K.. or more specifically, one that cuts off as completely as possible at or below Nyquist which would be 22.05K.

Look, start with the assumption that I actually know what I'm talking about.. then read what I said again.

[mitchiemasha]

Look, start with the assumption that I actually know what I'm talking about.. then read what I said again.

I've read what you wrote multiple times. I'm not ignoring any points, I'm studying them. The filter is covered in the video. I didn't doubt anywhere that you don't know what your talking about. I didn't realise we were disagreeing, just expanding.

As for this "The only signal possible at 20Khz is a sine wave." The only signal possible at any given frequency is a sine wave. Any other shape is a result of the mathematical relationship of the other sine waves.

All i stated was in my opinion, jumping to 24bits holds more benefit than increasing sample rate to 48k... But i suppose it depends on what one is doing. It could be debated, if wanting to slow down the audio, reducing the pitch, the 48k would have some extra benefit. However, the frequency content up there is so fast that shaped dither would seriously negate those benefits. The headroom 24 bit audio gets is well worth it. Especially a 32 bit float, it practically impossible to clip between insert with in a DAW. To which i only discovered a few weeks back, i was committing a serious crime of.

I am open to the debate and continually looking for a reason to start running my studio at 48k. When all my samples are 44k and I master to 44k or mp3. My CPU already strained.

[mitchiemasha]

I will add. This thread is finally getting on to the juicy stuff, content of worth, where we can learn, grow our understanding. I did go to rep your post on the noise floor, it was good information to highlight. But, apparently i have to spread it around some. I'm still undecided though. In an already noisy environment, surely any increase in dynamic range, the reduction of adding more noise, even if slight, is an improvement, especially under extreme amplification. (edit: but just to be clear, i have no problem with MP3's, I can't hear the difference, my ears are dead)!

[light-o-matic]

Yea 24 or even 32 bit is important in mixing applications where you are summing lots of signals and things are not at their optimum level yet. But 32 bit is useless for the final product, and 24 bit has only limited usefulness, because once the material is mastered it it going to be normalized up to digital FS, which means that the lowest signals are going to be significantly below the noise floor. I've listened to 24 bit vs 16 bit versions of the same material direct from producers I know and yea there is a difference, but in party environment you could never hear it.

But so far as the sampling rate issue, yea if you are shooting for a CD release then 44.1 is the way to go, because you'll have to resample anyway. But for WAV/FLAC/AIFF, 48 is going to give better results in a lot of cases because of the filter issue.

[mitchiemasha]

Ahhh Yes! Having the file at 24bit is next to pointless. What I've been referring to is the engine itself.

Edit: And when i say engine, i don't mean player, I mean DAW or DJ software.