Axiom Home Page Forums General Discussion Home Theater MP3 vs Lossless Compression - the proof

Now the MP3 guys and FLAC guys can weigh in against each other.

I made a quick little audio WAV file (sorry Mac guys - hope you can play this somehow) that shows the data that's missing from an MP3 that appears in an uncompressed or lossless compressed file.

I was actually surprised at how little detail is lost.

The MP3 was encoded with the Fraunhofer encoder for DOS (s-l-o-w but the best encoder I've heard - they started the whole push, so it made sense to use them as the standard) at 256kbps - which is a good balance of archival quality and file size, though some encoders do go higher, once you get past this bitrate you start getting into nearly the same file sizes as FLAC - why use a lossy codec if you can get a lossless for the same file size?

Sample for sample phase cancellation was used to produce a perfect copy of the differences between the two files, I have not adjusted the levels or "normalized" this file in any way, so it is quiet, you may have to turn up your speakers.

The sampled song used is New Model Army's "Queen of My Heart" - why? Was the closest thing I had at hand that's a pretty good mix of rock and electronic.

A 15 second clip - right click this link and choose Save As to save it to your local drive.

Bren R.

Bren, I would love to use FLAC, but there is no easy/inexpensive way to stream FLAC to my audio system from my PC.

Fraunhofer may have started it but they have lost faith with more then a few over there high charges for license fees for any of there A codecs and the speed as you have mentioned, is brutal to say the least. However Fraunhofer is by far the way to go if you rip at lower bitrates. LAME and a few others are designed with a higher bit rate in mind and are not user friendly for the most part. As you know I do it all at 320kbs which is overkill and I know it but when you have almost 300gigs of storage I don't lose any sleep over it.

As for Lossy Encoders vs. Lossless Compressors we can be here all day going over that one. In the end it comes down to which encoders make better decisions regarding what to keep and what to discard. One review that I read awhile back that stuck with me is this:

Downside of Lossless Codecs

"If lossless codecs are so good," you might be wondering, "why isn't everyone using them? Why do people even bother with MP3s and lossy codecs?" Good question. The bad thing about lossless codecs is that they offer poor file-size reduction .. compared to MP3s and lossy codecs.

In other words, a song compressed with Monkey's Audio or other lossless codecs will consume more hard drive space than the same song encoded with an MP3 or other lossy codec.

Lossless codecs typically reduces file sizes to half their original bulk. The exact amount depends on a number of factors, such as the type of music and the specific song. Live music doesn't compress well because of the background noise. Each song will be different.

High-quality MP3's typically require bit-rates of 175- to 225-kbps. Let's use 200-kbps as a happy median for purposes of making a comparison. Wave files from the original CDs are 1411-kbps. This number is calculated like so: 44.1 KHz * 2 channels * 16 bits = 1411.2-kbps [or nearly 10-MBytes/min].

Note that this calculation assumes k=1000-bytes. Most computers assume k=1024 bytes. So if your actual sizes differ slightly from what you calculate, this is why. Also the "b" in kbps = "bits" [not Bytes].

A 200-kbps MP3 file is ~14% the size [about 1/7th] of the original wave file [1411-kbps]. In other words, lossless files [typically 600-900 kbps] will be roughly 3 to 4 times larger than MP3s [consuming 3 to 4 times the hard disk space], but half as large as the original wave file ripped from the CD. You should now have a good feel for the pro's and con's of how lossy codecs compare with lossless compressors.

Seeing as how this thread will end up in some debate over compression we should go down the Dolby Digital vs. DTS road next!

All in favour say I! :-)

.....................and I have no real thoughts on VBR. I tried it, didn't like it and never went back. I should however try it again as it does make for a nice file size.

In reply to:

---

Bren, I would love to use FLAC, but there is no easy/inexpensive way to stream FLAC to my audio system from my PC.

---

Again, this is just to illustrate what is missing from an MP3 to drop the file size to 1/6 the size. Play what you will at home, this post is for educational purposes only.

Bren R.

BTW, Bren, WAVs are no problem for us Mac users. We can pretty much play any audio format... with the right software. I have a feeling I'd need extra software for FLAC, and I know I do for WMA and (hahahahaha) Real whatever it is, but WAVs...we got that covered.

In reply to:

---

Fraunhofer may have started it but they have lost faith with more then a few over there high charges for license fees for any of there A codecs and the speed as you have mentioned, is brutal to say the least. However Fraunhofer is by far the way to go if you rip at lower bitrates.

---

And if you have time on your hands, you get a darned good quality MP3 at the rate. As for it's speed, well, for someone like me, I compress a few songs a month for bands' websites as their new singles come out, and the difference between encoding at 1/5 real time and 4x real time is kind of moot. I'm more concerned about total quality for file size.

In reply to:

---

LAME and a few others are designed with a higher bit rate in mind and are not user friendly for the most part. As you know I do it all at 320kbs which is overkill and I know it but when you have almost 300gigs of storage I don't lose any sleep over it.

---

If you'd like to put your encoder where your mouth is, I'll gladly supply the same uncompressed 15 second WAV for you to compress at 320kbps with LAME and we can compare again.

Bren R.

In reply to:

---

If you'd like to put your encoder where your mouth is

---

To much coffee today? LAME can be had all over the net as can EAC. Go nut's! Post your graphs, post your notes, hell put up whatever you think you need to. At the end of all of this, I will still be using EAC and LAME. I have been having this debate with people since the dawn of MP3's and I'm not going to change now.

I'm a little lost as to where your going with this. Not once did I state that LAME in anyway shape or form was any better then Fraunhaffer. I just pointed out a few problems that Fraunhaffer has these day's. If you disagree with my comments on LAME being intended for a higher bitrate, then state why. Another one I forgot to mention is that they got sold to the people who brought us MP3PRO. Great idea, bad end result.

Well, a few things..

a) For those who're somewhat in disbelief that a 128k MP3 is not the 'CD-quality' it's advertized as, listen to this as a 128k MP3. Even at 320k, it has its flaws, with any MP3 encoder. I usually use this when people wonder why 128k MP3s off of kazaa don't satisfy me.

b) I'd suggest you try it with LAME and the --alt-presets. It's far better than the fraunhofer stuff now(except at very low bitrates), and it's been proven to be through ABX testing time and time again.

I find that lossy codecs don't make an immediate difference when encoded at high bitrates with well encoders, yet it does over a longer period of time. Up to 192k, if I listen enough, I can ABX 16/16 times the original from the MP3. Kind of tedious, but it was also nice to listen to music while on the computer. Sooner or later it becomes evident that something is missing, it doesn't sound as 'full'. With MP3, the higher frequencies sound somewhat blander and recessed at lower bitrates.

My Rio portable supports FLAC and Ogg Vorbis, so that is what I use. MP3 is a somewhat outdated codec, so if I choose a lossy encoder, I'd rather use something newer with a better psychoacoustic model.

Wow! This was exactly the kind of debate I was hoping for.

BrenR, you mentioned that you were surprised at how little was lost when you compared the differences in the two formats. To me that sounded like a lot, but maybe that's just because I have a hard time imagining what the song would sound like without those bits. When you play the two versions back to back, how noticable is the difference to you?

In reply to:

---

If you'd like to put your encoder where your mouth is
---------
To much coffee today?

---

Nah, it just sounds accusatory... doesn't have the same ring as "put your money where your mouth is"... that's all that was meant.

What I'm trying to do? It's purely for educational reasons... we know MP3s sound quite a bit like the original file, but what IS missing? What's the difference between them? Rather than using a bunch of flowery prose, I used some basic digital audio manipulation to actually hear the difference.

Everyone plays MP3s, but no one gives any thought to how they're compressed, and if you're not part of the team that designs the psychoacoustical model, you'll probably never understand what's rejected. Kind of cool to be able to see what's not in the compressed file.

Bren R.

In reply to:

---

b) I'd suggest you try it with LAME and the --alt-presets. It's far better than the fraunhofer stuff now(except at very low bitrates), and it's been proven to be through ABX testing time and time again.

---

Again, if anyone else has a copy of LAME and some time on their hands, I'll supply the original WAV and take back the MP3 and make a comparison. This wasn't meant as a "who's got the best MP3 encoder" thread, but if there's some interest, what the hell? Never hurts to learn... good discussion material at the next Star Trek convention, too!

In reply to:

---

Up to 192k, if I listen enough, I can ABX 16/16 times the original from the MP3.

---

Bro and I did the same testing between MP3 (his collection of downloaded material from Napster - as a practise, I don't download music, no... really - at most I have 50 MP3s - mostly deleted old punk vinyl and if Artificial Life want, I'll give them the $3 that their vinyl EP used to sell for)... back to the story, single blind A-B-C comparisons between MP3, CD-quality AIFF (raw rip of a CD) and "some Apple thing" (how I refer to anything on the Mac)... I could spot the MP3 (128-256kbps) every time, he played about 30 songs, but the "Apple thing" I couldn't distinguish from a full CD-quality AIFF. (I find out later the "Apple thing" is Apple's lossless codec. That cheater. )

But as I've mentioned before, I'm highly sensitive to digital artifacts, both in audio and video. Watching a BetaSX tape, especially speed-adjusted (slo-mo) drives me apesh*t.

Bren R.

Yeah, I've noticed "artifacts" in mp3's that I've listened to, but was never sure if it was just a problem with the encoding or a fundamental weakness of the standard itself. Of course, I've never listened to any that were encoded at better than 256kbs though, and most of them were 128 or so. Also, I've only listened to them in the car, on the computer, or through headphones. Never on a nice system, since I don't even have one... yet.

In reply to:

---

BrenR, you mentioned that you were surprised at how little was lost when you compared the differences in the two formats. To me that sounded like a lot, but maybe that's just because I have a hard time imagining what the song would sound like without those bits.

---

It may sound like a lot, but it's really only some harmonics and sibilance on the vocal track, the shaker and the drums, the instruments are nearly spot-on identical.

If I get some time I'll post an MP3 of that part of the original track for comparison.

As for side by side listening, they're close, but I can tell which is which, same as any MP3. Is the WAV 7x better like the file size would lead you to believe? Of course not.

Bren R.

In reply to:

---

Nah, it just sounds accusatory... doesn't have the same ring as "put your money where your mouth is"... that's all that was meant.

---

Sorry. I tend to get my guard up a little to fast around here these days!


All I know is this, if someone can show me a better way, I'm all for it! I would hate to go down the road of ripping everything again but if I have to, I'm in. Problem is I have gone down this road to many times in the past and nothing every beats what I do now in terms of quality, file size etc etc............at least to my ear. Is the MP3 format outdated?, dam straight it is but what do I turn to that will play on my computer, my Denon 2815 and the MP3 stacker I have in my van? lol That is one of my problems! I bought into this format long ago but nothing has come along yet that makes me want to switch.

If someone wants to say that a MP3 is technically flawed, you will get no argument out of me as I agree it's not perfect, but when it's done right, it isn't bad. MP3's at 160 and lower are not worth the time of day if you ask me. 192 is darn good and 256 has a little bit more going on then 192 but not by much. As I stated before, 320 is overkill and I know it but to me it's the truest reproduction I can get as far as a MP3 goes.

In reply to:

---

I've never listened to any that were encoded at better than 256kbs though, and most of them were 128 or so.

---

Say this with me 3 times.................128 is not CD quality! lol :-) Some putz made that statement years ago and it stuck. Most 128 and 160 MP3's that I have heard all seem to lack something.

I remember I ripped a track called Return Of the Mack by Mark Morrison. Not my type of music but the wife likes it and well It's a decent funky tune with lots of bass etc. Anyway, I ripped it at 128kbs and it sounded like crap. My sub was sitting there looking at me all confused! Had no real midrange, no nothing. I then ripped it at 160 and it was better but flat...............my sub was still feeling left out. Again at 192kbs and it came to life. Sub was feeling it but still not happy. Now this is the part I still don't get but is one of the main reasons I bought into 320kbs. I re-ripped it at 320kbs and my sub came alive! I did a check with my sound meter and everything and at 320kbs and at the same volume I was getting much better bass from my sub and my mains. Now this example is song specific and must have something to do with the original recording but I figure better safe then no bass!........................so everything gets done at 320kbs.

Ahhhhhhhhhhhh crap. Now I have that dam song in my head............................

Streem thread dejavuuuuuuuuuuuuuuu....................................

Return Of The Mack
Mark Morrison
Album: Return Of The Mack

Oh, oo-o-oh, come on, ooh, yeah
Well I tried to tell you so (yes, I did)
But I guess you didn't know, as I said the story goes
Baby, now I got the flow
'Cos I knew it from the start
Baby, when you broke my heart
That I had to come again, and show you that I'm real

(You lied to me) all those times I said that I love you
(You lied to me) yes, I tried, yes, I tried
(You lied to me) even though you know I'd die for you
(You lied to me) yes, I cried, yes, I cried

1-(Return of the Mack) it is
(Return of the Mack) come on
(Return of the Mack) oh my God
(You know that I'll be back) here I am
(Return of the Mack) once again
(Return of the Mack) pump up the world
(Return of the Mack) watch my flow
(You know that I'll be back) here I go

So I'm back up in the game
Running things to keep my swing
Letting all the people know
That I'm back to run the show
'Cos what you did, you know, was wrong
And all the nasty things you've done
So, baby, listen carefully
While I sing my come-back song

2-(You lied to me) 'cos she said she'd never turn on me
(You lied to me) but you did, but you do
(You lied to me) all these pains you said I'd never feel
(You lied to me) but I do, but I do, do, do

(Return of the Mack) here it is
(Return of the Mack) hold on
(Return of the Mack) don't you know
(You know that I'll be back) here I go
(Return of the Mack) oh little girl
(Return of the Mack) wants my pearl
(Return of the Mack) up and down
(You know that I'll be back) round and round
(rpt 2, 1)

(You know that I'll be back) don't you know

(Return of the Mack) here it is
(Return of the Mack) hold on
(Return of the Mack) be strong
(You know that I'll be back) here I go
(Return of the Mack) my little girl
(Return of the Mack) wants my pearl
(Return of the Mack) up and down
(You know that I'll be back) round and round

In reply to:

---

Sorry. I tend to get my guard up a little to fast around here these days!
(snip)
If someone wants to say that a MP3 is technically flawed, you will get no argument out of me as I agree it's not perfect, but when it's done right, it isn't bad.

---

The funny part is, we're arguing the same point from different angles...

"No, you idiot, let's attack the lizardmen with daggers!"
'No, you fool... I say we slay them with arrows'
*fight erupts, Lizardmen fall over their treasure laughing*

Again, I was surprised at how little is lost in the translation. MP3 is outdated (look how fast it came out on the tail of MP2s) and now that it's almost a trade name and has an installed base of, like, 8 billion players, there have been no major changes to it.

Bren R.

The "Apple thing" is called aac. I've found it to sound much much better than an mp3 (both at 192kbps).

One mans junk is another................ah you know the rest! lol

In reply to:

---

Conclusions
If you care about audio quality and listen to a lot of electronic music, avoid MPEG-4 AAC, particularly if you use bit rates below 160kbps. iTunes MP3 at 160kbps is better than AAC at 128kbps if you can afford the extra file size. If you really want CD quality, you'll need to use Ogg Vorbis or LAME MP3 encoding.

---

http://www.xciv.org/~meta/audio-shootout/

This is what is fun about this stuff. For every 10 people that say AAC sounds better, 10 more will say it doesn't.

This is another good one to read (Although it's tough to figure out what the hell he actualy likes!):

http://www.recordstorereview.com/misc/aacmp3.shtml

Guys, I'm fascinated by this discussion. And I continue to be grateful for the generous sharing of knowledge and experience that takes place here.

And you've got to either speak slower, or send me to the "Ripping all those CD's to a Hard Disk for Dummies" page.

Seriously, I admire the scienctific and experienced way you approach this question. Problem is, I'm looking for a consumer-based recommendation. Right now, I'm still loading CD's into a carousel player that won't die. Sounds pretty good.

But, like most, I am seduced by the power of an uber-jukebox. If I were to invest the requisite time, money and energy into ripping my entire collection to a HTPC-type-device, I'd need real-world help. And I'm more confused that ever at the moment.

Here are my criteria:
1. Sound quality virtually indistinguishable from the CD
2. File sizes small enough to get 400-500 CD's on readily available, competitively-priced hardware
3. Ripping speeds appreciably faster than real time.
4. Ripping and Music library management software that is easy to use and compliant with listener-area-controllers
5. No super geeky or extremely obscure software/hardware solutions. I am not the alpha geek that most of you are.

I somehow get the impression that the LAME solution is slow and/or obscure. Is that correct?

Is just using something like Musicmatch Jukebox and cranking the bitrate up to 320 going to get me the same kind of results?

Be gentle.

Thanks for your guidance.

Tom,

I use CDex to rip CDs...take a few minutes per CD. It uses the LAME engine. It can also rip to other formats as well.

I have a 250gb drive for storage in my PC that cost me $120. That size of drive, at a rip bitrate of 256, should hold close to 2500 CDs.

For library management, I like iTunes....but there are plenty of others.

I also use CDeX for straight audio rips... as discussed, I use the Fraunhofer engine separately to compress MP3s, but CDeX will rip (calling the LAME command line) directly to MP3.

Bren R.

I was just about to post a message EXACTLY like Tom...

I beginning to like this idea, but all this mumbo jumbo about ripping cd's a million different ways is making me light headed.

spiff....the process is extremely simple.

I'm sure it is. I just want to be sure to get the highest quality rips I can. I'd rather get two drives, then be listening to something of lesser quality than I could be with my cds.

Bren and Curtis, thank you both very much. I appreciate the clarification.

Craig, I was thinking of you, and hoping that I was not the only one who got a little lost.

Keep us posted on your project.

spiff...you really need to come over and listen. Bring one of your CDs, we will play it and rip it, then you can compare.

Why two drives?

Just suggesting that I'd rather buy two drives, if I need to rip them at a file size that is rather large to retain the quality, than go cheap, use only one drive and rip at a smaller file size and suffer a quality loss. Make sense?

The largest bitrate for MP3 I believe is 320....so even at that, with a 250gb drive, you are looking at close to 2000 CDs.

320kbs is in fact the max at this point in time.

In reply to:

---

Here are my criteria:
1. Sound quality virtually indistinguishable from the CD
2. File sizes small enough to get 400-500 CD's on readily available, competitively-priced hardware
3. Ripping speeds appreciably faster than real time.
4. Ripping and Music library management software that is easy to use and compliant with listener-area-controllers
5. No super geeky or extremely obscure software/hardware solutions. I am not the alpha geek that most of you are.

---

1 - That can be done even with a MP3 at 192kbs if it's done right......I prefer 320kbs. Better safe then no Bass! lol

2 - Already out there!

3 - Speed can be a two way street. Rip to fast and you can get read errors and a few other problems. One thing with EAC and LAME is it's not the fastest way to go. I can use MusicMatch and rip a CD in under 3 minutes but it does not sound the same as when I do it with EAC and LAME. Some do but not all so I take my time and do it the way I think it's right the first time. My method however can take upwards of 10 minutes per.

4 - Tons and tons out there..............you know what and how I rip. As for how I look after my library, Winamp!

5 - It's not the hard once you decide on what software you want use and what codec it is you prefer.

As for LAME being slow and obscure. Not at all. LAME is all over the place and very popular. As for the speed, it all comes down to what software you use, what codec and how good is your CD drive that you want to rip from.

Does that help at all?

Yes, John, it does. Thank you very much for taking the time to explain it to me. I appreciate you sharing your knowledge and experience with those of us who are new to that process.

I'll point out that if you're going to use iTunes for management (and you're not streaming through something that requires MP3s) you can rip with iTunes, using AAC with a high bitrate (160 or above-it's different for AAC) or with the Apple Lossless (assuming you've got tons of disk space).

Oops, I just made it more complicated, didn't I?

Anyway, you don't need a separate program to rip your stuff. I've read a few articles showing AAC as ripped by iTunes to be higher quality than other formats, but I don't have the links, and I've also read articles that say different.

Great discussion.

I'm currently going through the process of ripping my CD collection to MP3. I'm using Exact Audio Copy (EAC) with the LAME encoder running variable bit rate (VBR). The detailed settings are modeled after the suggestions here

I've been ripping mp3s fast and furious for the past few weeks. I've only listened to a small subset of what I've ripped so far, but I've been extremely pleased with what I've heard.

I have noticed that the compression logs often show "timing problems" occurred during the process, but ultimately there are no errors and I haven't heard any adverse effects in the mp3s themselves. Can anyone explain what these timing problems are? How might they affect the sound of the mp3s if at all, and is there any way or need to try to avoid them?

Thanks in advance!
Scott

In reply to:

---

I have noticed that the compression logs often show "timing problems" occurred during the process, but ultimately there are no errors and I haven't heard any adverse effects in the mp3s themselves. Can anyone explain what these timing problems are? How might they affect the sound of the mp3s if at all, and is there any way or need to try to avoid them?

---

Jitter... ie: the ripper is Lucy and Ethel boxing chocolates in that famous I Love Lucy episode. It missed a chocolate.

Slow down the speed you rip at, use a bigger buffer size - bunch of different ways to keep it from happening, but if you're just going to MP3 anyway and you don't hear a little noise burst anywhere, meh... you can probably get away without re-ripping the track.

You'd have to check with EAC itself to find out whether it passes the error along, or checks with the Reed-Solomon coding to try to fix the error on the hardware side, or forces a re-read of the affected sectors... seems a pretty generic error message.

Chances are you're just ripping too fast... if the drive is pushing data at say 32x single spin speed, it's got to spin-up and slow down, spin-up and slow down to let the encoder catch up. Like running a marathon by sprinting and resting.

Bren R.

Exactly! Sounds like a speed issue more then anything else. Are you using Secure mode in EAC? I do and my max rip speed is around the 3X mark. Slow but accurate.

EAC will also take it's time going back over the track to look for errors etc. It all adds time to the process but worth it in the long run.

I'll double-check when I get home, but I know I was set up in Secure mode when I got started. Unless something was changed without my knowing, it should still be that way.

From the little extra reading I've done, it seems that these "timing problems" or jitter do not necessarily contribute negatively to the end product. It's just an indication that the CD or DVD drive took longer to respond than EAC expected. Since I haven't heard any issues in the MP3s I've listened to, I think I'm going to keep plugging along.

I'll have to admit, I've probably been a little greedy. I discovered that I could rip from both my CD-R drive and DVD drive at the same time. I know that the extra load on the machine might not be desirable, but the time saved and the seemingly good end result kept me going.

Not to mention, I'm about 75-80% of the way through roughly 400 cds. The thought of trying this again with a different method is just plain scary.

If there's one thing I know, it's my digital audio theory.

After seeing Digitize Design Group's (a Finnish C64 "demo" group) demo - Sabrina "Boys, Boys, Boys" (available for the C64 emulator users here) in 1989, I HAD to learn about this way to get real music to play on my computer... none of this SID chip FM-synthesized stuff.

Long story short - I built a digitizer and started messing around with compression (48K isn't a lot to work with), mixing and sequencing.

Then came the Sound Blaster (8 bit mono? Beats 4 bit mono!) and the PC and it was a whole new ballgame.

Bren R.

Ok, this is going to sound like a plug here, and I guess that it is since I know the guy, but what the heck. A few folks in the thread have mentioned that they're intimidated by trying to rip all their CD's themselves. I admit that it's not difficult to do this for a few CD's but when you have a bunch, the process can get a little daunting. Anyway, a friend of mine is in the process of starting a business where for a few bucks he'll rip your CD's for you in most any format you'd like and then burn them to one or more DVDs for you. I was planning on having him rip my 200 or so CD's and then just copy the DVD(s) to my hard drive. Then I'd proceed with the listening to the ripped songs using the previously mention SqeezeBox or AudioTron. He hasn't actually performed this service for any paying customers yet, but I'm sure he'd appreciate any feedback any of you had on his concept. You can check out his site at FreeYourCDs.

The free alternative, of course, is to do it yourself over time. Most ripping programs I am aware of will automatically contact the CDDB and get the album information for each CD, saving you from having to manually enter any data. They can also be set to auto-rip mode so all you have to do is drop in a CD. It'll rip it and then eject it so you can drop in the next CD. Keep going until you get sick of it.

You can burn your entire library piecemeal in less time than you think.

Hi pmbuko,

Yes, I whole heartedly agree with you. The cheapest way of doing this is to do it yourself. You can do it for free and can get the results you want that way. But it will take time - if you have the time that's fine.

The service I'm in the process of starting up (apologies for the still lame web site: www.fycd.com) is all about convenience. I first got interested in the idea when after three years of having an MP3 player I still found that:
(a) I didn't have my enitre collection copied (I only had about 25% done - which represented my current favorites)
(b) I had a variety of formats and still hadn't decided what was best (the answer is, of course, one size doesn't fit all!)

I searched for the idea of a "bulk encoding service" but couldn't find anything at the time. So I plunked down a couple of grand on some equipment and got busy designing, coding and modifying. While doing this I've since found about half a dozen competitors (but I think I'm better than them - of course you need to decide for yourself).

What makes the FYCD service unique is the widest range of encoding / compression choices and also a unique "DIY" option. With the "DIY" option we supply you with your CDs copied to WAVs and some software (just Windows at this point - but I'll tackle others later) to allow bulk encoding/compression to your choice. This approach allows you to (a) encode to different formats as your needs change (b) re-encode in the future when a new encoder or compressor hits the streets.

This is still very much in "beta mode" but I'd love to hear from any and all on whether this is a great idea or a lousy idea.

Many thanks if you took the time to read this. Sorry if I'm breaking ettiquette by posting a plug - but if you help me out then I'll try and help you out with the FYCD service

I didn't mean to knock your service. I'm sure it would be worth a lot of people's time to take care of it all at once. I just wanted to make sure people were aware of the DIY route.

Hey, just popping in for a bit.

Bren, interesting thread. However, I'm not sure of the methodology in your experiment- hear me out.

***Warning: somewhat technical***

MP3, like all modern compressed audio formats, relies on the imperfections of the human ear, significantly in the phenomenon of 'masking'. That is, a tone of a certain frequency may 'mask' weaker tones around it. Think of an MP3 encoder calculating a spectrum graph every 1/75 of a second, finding the spectrum peaks, and marking all nearby weaker tones as 'masked' (the ear ignores these in the presence of the strong tone). The MP3 codec throws these masked tones out.

Fast forward to doing a waveform difference on compressed vs uncompressed audio: part of the difference will be these 'completely masked' tones, but as the dominant, masking tone is gone (since it's the same), you'll hear these otherwise-invisible sounds.

Likewise, if the psychoacoustic model MP3 uses is wrong for your ears, just a little difference in the waveform may unmask borderline tones or mask otherwise-audible tones.

So, neither of the two 'obvious' conditions hold: either that a small difference in waveforms strongly implies that the two waveforms will sound the same to a given person, or that a big difference in waveforms strongly implies that the two waveforms will sound different to a given person.

Counter-intuitive, I know. Blame our ears.

RD

In reply to:

---

Fast forward to doing a waveform difference on compressed vs uncompressed audio: part of the difference will be these 'completely masked' tones, but as the dominant, masking tone is gone (since it's the same), you'll hear these otherwise-invisible sounds.

---

Exactly - part of the MP3 compression scheme they claim is that if a triangle is being struck at the same time as a bass drum, you'll never hear the triangle - but how does a computerized encoder know what's a triangle and what's a bass drum? I'm assuming that if there's a loud sound in a certain frequency band, that it ignores or uses less resolution in the other frequencies. It's hard to say. In the end, I was just showing what the differences were between the two. People can make their own decisions.

In reply to:

---

Likewise, if the psychoacoustic model MP3 uses is wrong for your ears, just a little difference in the waveform may unmask borderline tones or mask otherwise-audible tones.

---

You speak of the Golden-Earred?

Bren R.

In reply to:

---

You speak of the Golden-Earred?

---

Didn't they do "Radar Love" or some other 70's arena rock anthem?

Hey, if CDex calls the LAME encoder, what's the difference between using that and using EAC? Is that a stupid question?

Also, Neverhappy really likes the 320 bit coding, and the quality-preservationist in me buys that premise. Is there ANY advantage (besides saving some disk space) of using variable bit rate instead? Is there a circumstance in which VBR could be better than 320? Why don't some people like using VBR?

Thanks for suffering my questions.

In reply to:

---

Is there ANY advantage (besides saving some disk space) of using variable bit rate instead?

---

Other then space, I can think of none.

In reply to:

---

Is there a circumstance in which VBR could be better than 320?

---

See above! The main reason I don't like VBR is simple. It's yet one more thing your letting the codec decide. How can it know what you want left in and what you want taken out? So I keep it at 320 and take it out of the equation.

In reply to:

---

Why don't some people like using VBR?

---

The question should be, why do some people use it? lol Just kidding on that one!

In reply to:

---

Hey, if CDex calls the LAME encoder, what's the difference between using that and using EAC? Is that a stupid question?

---

All ripping software is not the same and I will leave it at that. The way they look for errors to the way they actually handle errors is very different from software to software.

In reply to:

---

Didn't they do "Radar Love" or some other 70's arena rock anthem?

---

And Twilight Zone!

In reply to:

---

Hey, if CDex calls the LAME encoder, what's the difference between using that and using EAC? Is that a stupid question?

---

One's even got the word Exact in it's name... it's gotta be good!

In reply to:

---

Also, Neverhappy really likes the 320 bit coding, and the quality-preservationist in me buys that premise. Is there ANY advantage (besides saving some disk space) of using variable bit rate instead?

---

VBR saves some space (for downloading off websites or what-have-you) as for not liking using VBR - you're already asking a codec to decide what you can and can't hear, might as well loosen the leash and let it decide to what extent you can and can't hear as well!

Bren R.

Thanks again, John. I appreciate your time and patience.

Aye, a lot of good information here.

To chip in regarding EAC vs CDex, they both do some pretty advanced things to minimize jitter, etc, but EAC has what is called "Secure Mode"* where it reads each part of a CD at least twice and compares the result-- if it doesn't get the same both times, it carefully reads that part over and over to get the correct result. If that fails, it'll tell you. I find it a bit slower than other rippers because of this, but it's worth the peace of mind to me. There are a few more features of EAC that I don't believe CDex has, but that's the major one.

*I'm almost sure that CDex doesn't do this, and I checked the webpage, but it's quite possible that I'm wrong.

Regarding VBR vs constant bitrate, I agree with NH that there's no quality gain from going VBR over just going 320k. However, for, say, a 192k file, VBR will be hugely better. Another way to look at the debate is "constant bitrate is variable quality, variable bitrate is constant quality". VBR got a bad reputation because of some early bugs in MP3 encoders and players, but all that's over now. It's good stuff, especially LAME's implimentation.

Reply to Bren following soon.

When setting up a VBR encoding, the bit-rate you choose is the guaranteed minimum, right? For example, if I set up an encoding session, pick a root bitrate of 192 and enable VBR, that means I'll have a guaranteed minimum bit-rate of 192 in all my songs, but the encoder will boost it higher as necessary to get a cleaner sound.

At first, I thought VBR meant the encoder could drop OR raise the bitrate.

"Exactly - part of the MP3 compression scheme they claim is that if a triangle is being struck at the same time as a bass drum, you'll never hear the triangle - but how does a computerized encoder know what's a triangle and what's a bass drum? I'm assuming that if there's a loud sound in a certain frequency band, that it ignores or uses less resolution in the other frequencies. It's hard to say. In the end, I was just showing what the differences were between the two. People can make their own decisions."

How the MP3 codec calculates masking is actually pretty cool-- imagine making a spectrum graph 75 times per second, finding the peaks, then drawing a diagonal line down both directions from each peak (the slope of the line depends on various and sundry things). Anything below the line is masked (this is the core of how masking works, though there's much more math behind it).

Depending on the harmonics and tuning of the bass drum and triangle, the triangle will probably be masked-- but the beauty of this method is that the encoder can just rely on this 'geometric' method and not know what's making the sounds.

"You speak of the Golden-Earred? "

Perhaps. The psychoacoustic model in LAME is wonderful, but psychoacoustic models aren't always one-size-fits-all.

RD

PMB, that really depends on the program.

Technically speaking, the VBR bit-rate you choose is supposed to be the guaranteed minimum; however, it's rather ambiguous whether some programs use standard VBR when you choose VBR, or a type of VBR called ABR ("average bitrate"), which would aim for an average bitrate centered on the bitrate you chose.

It's worth checking to see if your program makes a distinction between VBR and ABR.

By the way- a neat thing about VBR is that, when there's only absolute silence, the bitrate can drop down to 32k (even if you set a hard minimum). With 320k, you use 320k for silence, too.

RD

In reply to:
Is there ANY advantage (besides saving some disk space) of using variable bit rate instead?


One potential issue (other than saving some disk space) could be the network bandwidth required to feed the mp3's from a PC to a remote digital video player. Now, if all of your components are hardwired, or you have a solid 802.11g network, I'm sure you'll be fine streaming whole uncompressed .wav files. If you have a marginial 802.11b network? Who knows?

I used 802.11b to stream my MP3's(256kbs no VBR) with no issues.....surf the web with my laptop at the same time too.

Ok, maybe I was worring about nothing. I checked around a little and it looks like a typical real world 802.11b connection running wireless encryption (as you should), will get at least 2Mbps (probably more). A 320kbps MP3 will only use about 1/6th of the connections throughput -- plenty of room left for surfing.

Now here's the confusing part to me. By my estimates, a fully uncompressed wav file is only around 170kbps? Huh? I must have made a mistake somewhere. Here are my assumptions:
A typical 20 minute CD uses about 1/3rd of a 600MB CD, therefore 20 minutes of wav files should be around 200MB. This yields around 204,800,000 bytes in 20 minutes. Or about 170,000 bps = 170kbs. That doesn't seem right, that would mean that a high quality MP3 is using almost twice the bandwidth of a raw wav file. There's obviously something I'm missing here.


Btw, here's a link that listed some "real world" wireless connection bandwidth info...
http://www.homenethelp.com/802.11b/index.asp

The mistake you are making is in units. You have to understand that 8 bits (binary digits) make up one byte (binary term). Redbook CD Audio at 44.1 kHz and with a resolution of 16 bits takes 176.375 KB/s (KBps or KiloByte / sec) or 1411 Kbps (Kb/s or Kilobit/sec) (which is significantly higher than a compressed MP3, even at 320 Kbps.)

As a side note: transfer rates for network connections are generally reported in base 10 instead of base 2 and are Kbps (not KB/s).

Doh! I can't believe I made that mistake. Thanks for clearing that up. So that means that the MP3 would be using 1/8th of 1/6th (in round terms) of the bandwith of an average slow wireless network connection. That's something like only 2%.

Well let's assume that an 11 Mbps wireless network gets 6.4 Mbps real-world transfer rate. This would be 6400 Kbps (not [6.4 * 1024] since network transfer rates are base 10 and not base 2), which equates to twenty times the bandwidth required by a 320 Kbps MP3. In this scenario, the MP3 would take 1/20, or 5 % of the wireless bandwidth. In the real world, unless you have a very good transmitter and card (or a signal amplifier somewhere in the chain), you are not going to achieve this speed at a long distance. If I walk around my apartment complex into friends' places, even with my 802.11 G network I often dip down to about 1 Mbps which is 1000 Kbps. This would mean that if I am playing a 320 Kbps MP3, I would be using roughly 1/3 of the network bandwidth.

(Note: I am not taking into consideration any error correction and which would lower the transfer speed further when the signal is poor.)

Also keep in mind, that at this speed (1000 Kbps), we are under the 1411 Kbps requirement for Redbook Audio CD Playback. Here is a situation in which compression proves useful.

The math:

Uncompressed redbook audio:
44100 (samples) x 16 (bits - 2 bytes) x 2 (channels) = 1411200 bps or / 1024 = 1378 Kbps

Bren R.

BrenR's correct... Just that since it's in bits instead of bytes, it is not 1378, but rather 1411.2 Kbps (since throughput is base 10 number system not base 2).

In reply to:

---

Just that since it's in bits instead of bytes, it is not 1378, but rather 1411.2 Kbps (since throughput is base 10 number system not base 2).

---

Oops... yes, it's kilo as in the metric prefix for thousands. I originally had it in KB/s, of which there are 1024 bytes to a KB (and 1024 KB to a MB, 1024 MB to a GB - regardless of what the snot-nose at Future Shop tells you), realized my answer was in a different unit and forgot to change the equation.

Bren R.

Yep yep - I'm still surprised at how many people think that a KB is 1000 Bytes... Scary how they end up working for computer shops, heh.

The really sad thing is the whole disk size "scandal." Well, I call it a scandal. I don't understand why disk sizes are measured base 10, so that a "120 GB" drive is really only ~111.76 GB. Then formatting and partitioning it gives us even LESS usable space; simply unfair!

In reply to:

---

I don't understand why disk sizes are measured base 10, so that a "120 GB" drive is really only ~111.76 GB.

---

I think the first time I ran into that was with my gigantic Samsung I think it was... 560MB or so... was actually 546 or so, then after an FDISK and FORMAT... less again.

I felt ripped off.

Bren R.

Such a scam! Here's another computer scam that upsets me: CAV (constant angular velocity) drives only giving max speed, instead of both the minimum and maximum. If everything were just CLV (constant linear velocity), there would be no problems with this, but every drive maker wants to have the highest speed rating - *shrug*.

Very interesting, Bren and Will, but true audiophiles want to know which sounds better, a bit or a byte?

A byte - it would contain 8x the information.

Ha - bet you didn't expect a clear answer.

Bren R.

Technically speaking, one bit would be a square wave... Have you ever heard one of those? I hope you haven't put your speakers (especially the tweeters) through that kind of evil punishment!! One byte would be equivalent to a 4 bt sound source if it's in stereo or an 8 bit mono signal -- talk about a limited dynamic range!

In reply to:

---

Technically speaking, one bit would be a square wave... Have you ever heard one of those?

---

But of course - I've programmed the MOS 6581/8580 ICs.

Hmm... ring modulation - there's a term I haven't heard in so long.

Bren R.

Okay, so then a byte would sound a bit(or rather 7 bits)better.

Semi-seriously though, 8 bits would allow for about a 50dB dynamic range, which is a heck of a lot better than most of the pop stuff coming out these days.

Don't forget about a word and a nibble (my personal fav)

Spelled nybble (4 bits).

Bren R.