I've been RIPping stuff for a while now in preparation for the empeg; I'm using Audio Catalyst.

I am producing Variable bitrate MP3s @ 44.1; on playback through WinAmp I am noticing that there is some distortion on deep bass notes. Any comments on settings/fixes or explanations? Anyone else seeing similar effects?

Uh Oh. I kind of thought that this (or something similar) might happen with MP3 compression. I compete in IASCA, so sound quality isn't important to me -- it's EVERYTHING. This is why I am planning on storing my music in MP2 compression, rather than MP3. It uses twice as much space, but from what people tell me, there is no discernable (or even measurable) loss of quality when comparing an MP2 compressed file to the original. The radio station where I work airs all of its music from hard drive, and it is all compressed MP2. empeg has told me that their player will play MP2 and MP3 files indiscriminately - you can mix and match them and the player will automatically uncompress and play them.

tanstaafl.

"There Ain't No Such Thing As A Free Lunch"

Yeah, I followed your posts. It looks like I need to do a back to back comparison between the original track on CD and the MP3 through the same audio path, with a pair of borrowed Sennheisers

So tell me, how do you rip to MP2?

Jo-Jim says hello.

Tanstaafl,

can you send me your email? I wanted to ask you a few questions about my stereo i'm putting together. I've been to an IASCA competation, and i also believe the sound quality is EVERYTHING!

Thanks for the info.

My email is: [email protected]
thanks.



-CHiP

I just went back and checked all my e-mails from Hugo and the boys at empeg. He did state that the empeg supports both MP2 and MP3 compression and that you can mix and match them indiscriminately, and while he kind of implied that the supplied empeg software would RIP MP2 as well as MP3, he didn't come right out and say it. Since I am #8741 in the waiting queue, I figure I have a lot of time available to resolve this difficulty!

tanstaafl.

"There Ain't No Such Thing As A Free Lunch"

No, there's no ripping s/w with the unit. It came with a playlist and download manager application, but you have to get your music ripped yourself.

Keep in mind it also might be because of his encoder. I've not noticed any problems with bass or even trouble (Mp3 compression can sometimes distort treble slightly however) on most of my Mp3's, both that I have made, and that I have recieved from others. I use the Fraunhoffer codec on High Quality, works great. From everything that I've read and from my own tests, this is the highest quality MP3 encoder. It is however definately slower than other implimentations. So keep in mind with quality comes a drop in speed. I suggest doing your own tests with Mp3 and Mp2 files for qualtiy, don't assume that Mp3 isn't a good option based on one persons experience, especially with a lack of further testing. Anyways, just my 2 cents. :)

- JavaJones

I was under the impression that Layer 3 produced a higher quality output for the same bit rate, at the expense of encoder complexity.
In my experience a 256 Kb/s MP3 is indistinguishable from a CD when played through good quality hi-fi equipment.
I would recommend trying (as I have done) playing a 128 Kb MP3, a 128Kb MP2, the ripped WAV and the CD track from Winamp so that you can compare and switch quickly between them.

MP3 seems to lose quality regardless of the bitrate used, which is probably due to the acoustic masking algorithm it uses (which doesn't just remove inaudible frequencies most of the time). Given the right equipment you can still tell the difference with 256kbit layer 3 encodes. As for layer 2, it's unfair to compare it at 128kbit as it wasn't intended for bitrates that low. 256kbit would be a fair test.


- John.

As for layer 2, it's unfair to compare it at 128kbit as it wasn't intended for bitrates that low. 256kbit would be a fair test.

I was under the impression the layer 2 standard supported bitrates from 32 to 384 kbit/s. [MPEG Audio]
Are you saying that although 128kbit/s is supported it is not meant to be used?


Given the right equipment you can still tell the difference with 256kbit layer 3 encodes.

I agree that this is a possibility, I personally can't tell the difference... but the point was that for the same bitrate layer 3 should produce higher quality audio.


As a matter of interest, have you actually tried attempting to distinguish between 256kbit/s layer 2/3 and the original CD? If so, any specific track? I'd like to try it out myself.

Bryan

While layer 2 supports bitrates down to 32kbit, layer 3 sounds better at low bitrates. Layer 2 sounds fine at 128kbit so long as you don't give it anything musical to deal with.

It's been a long time since I tried layer 2 compression (years), I should try again with the latest codecs to see if anything's improved. At high bitrates layer 2 should be better than layer 3 simply because less is being done to the audio. Layer 3 supposedly only removes audio "you can't hear", but frankly that isn't quite true. I may be wrong, but I think video cd actually uses layer 2 - and I can't hear anything wrong with the audio on video cd.


- John.

(The above may not represent the views of Empeg :)

At high bitrates layer 2 should be better than layer 3 simply because less is being done to the audio.

I don’t understand how this follows. For any given bit rate they both have to compress the same amount of information into the same space.
My understanding (or lack thereof) was that they are both perceptual encoding (lossy) only the complexity was different. i.e. Layer 3 tries harder to cram data into the bit stream using a better perceptual model. Thus produces higher quality output.

As an example let’s assume say a 256k bit stream. This means that from the original (CD ~1400Kb/s) a compression of about 5:1 is required regardless of the method (Layer 2 or 3). Layer 2 does less to the audio i.e. it doesn’t try very hard when throwing away some of the information. While layer 3 has a better model and tries harder to decide which sections of the information can be thrown away.
The same amount of information is thrown away but layer 3 is better at deciding what to remove?



That started me thinking, is 5:1 loss-less audio compression possible? (Sounds hard, but not that hard...)


More musing... What if you removed the restriction of streaming and compressed globally (on a track basis)?


Bryan

If the layer 3 implementations were done perfectly, a 256kbit layer 3 stream would sound better than a 256kbit layer 2 stream - but it's not done perfectly. It's nearly impossible to say for certain whether the ear can hear a frequency or not without massive amounts of calculation, which just isn't done because it would take forever to compress.

I doubt you could get 5:1 lossless compression, maybe 2:1 at the most. On completely random data you can't lossless compress anyway. Sound happens to be quite a random signal.

Not sure what you mean by 'compressing globally'.

- John.

(The above may not represent the views of Empeg :)

Not sure what you mean by 'compressing globally'.

I must admit I don’t know squat about information theory or high powered maths. It just seemed intuitive to me that if you considered the file as a whole and then attempted to find redundant (compressible) sections, you could do nearly as well as considering frames that you have to squash into a certain bit rate.

I guess I should have a play with this compression stuff myself to try and understand it.

Bryan

It's perfectly valid to compress the file as a whole by converting it in one huge transform, finding redundant frequencies, and compressing. Course, the decoder would have to do a huge inverse transform to get it back again - that means lots of RAM. It would actually be faster doing a single large transform though.

While it sounds like a great idea, giving you the best statistical compression over the whole, this isn't ideal because you get bad compression over short periods of highly detailed audio where you would want different frequencies removed than other periods. It's a bit of a trade-off finding the best block size for that sort of thing. MP3 uses 1152 sample blocks for compression, which translates to about 1/38th of a second, and probably isnt a bad choice (1024 would have been nicer though :)


- John.

(The above may not represent the views of Empeg :)

Hmm there's another subject which I don't believe you have touched.. the software used to compress the MP3/MP2 files makes a huge difference.. encoding with Xing(Audiocatalyst) is certanly faster than the old mp3end from Frauenhof, but the quality if also a lot worse.. get BladeEnc/Radium Coded and compare the result to Xing, you don't even need a good hifi to hear the difference.

DrRadiaki - waiting in patience for the empeg :-)

Ages back I tried comparing the quality of the fraunhoffer and xing codecs. The old xing codec had problems with general noise from the low quality algorithm, and despite Xing claiming they filtered all frequencies above 16kHz, they still got through somewhat, and the filter itself caused lots of distortion to frequencies just at the limit. No such problems with fraunhoffer. Removing high frequencies is cheating, anyway - that's just throwing data away which you want compressing. If you really don't want high frequencies, use 22 or 36kHz sampling rates.

I also came across very audible problems with a version of BladeEnc - not sure which one because I haven't managed to reproduce it myself. It would encode fine, but it would 'ring' loudly at around 15kHz - and then only in the left channel. Quite obviously a bug, which has hopefully since been fixed. It's also a bloody slow algorithm. My best efforts at optimizing it have come up with a 25% speed increase, but I don't see it getting any faster without the author rethinking some of the dreadfully slow operations performed (e.g arc tangents).

The new Xing codec in audiocatalyst sounds pretty much fine to me. I should try to analyse the results of the new codecs at some point to get a more quantitative picture of which is best.

- John.

(The above may not represent the views of Empeg :)

>Hmm there's another subject which I don't believe you have touched.. the software used to compress the MP3/MP2 files makes a huge
>difference.. encoding with Xing(Audiocatalyst) is certanly faster than the old mp3end from Frauenhof, but the quality if also a lot worse..
>get BladeEnc/Radium Coded and compare the result to Xing, you don't even need a good hifi to hear the difference.

Actually, I found bladeenc inferior to AudioCatalyst VBR files. There were a lot more artifacts present in the bladeenc output. I believe the VBR factor makes a big difference in terms of SQ versus file size - the VBR were about the same size but much better SQ.

Just to add another opinion here, my experience is that the Fraunhofer encoder is the best at a fixed bit rate, with the Xing and then the BladeEnc following behind. At low bit rates (32K, 24K), the difference is obvious (as to why I'm encoding at those rates, that's another story). Xing's VBR encoding is better for a particular size of file because more complex blocks are being encoded at higher bit rates. MPEG audio was in part designed for specific bandwidth streams where it's easier to guarantee a certain speed of transfer rather than deal with lumps of high speed and lumps of low.

Feed the Hungry. Save the Whales. Free the Mallocs.

Would anybody with MP3 experience like to go out on a limb here, and recommend which encoder is best given the following priorities: I don't care in the least how long the encoding takes; I am only marginally concerned about file sizes (I'll have either 10, 18, 0r 20GB to play with, depending...); I will pay nearly any cost or penalty to maximize sound quality.

At this point I have had no experience with MP3. Do I have a big learning curve ahead of me? Am I doomed to disappointment?

tanstaafl.

"There Ain't No Such Thing As A Free Lunch"

The learning curve is not especially steep, merely expensive

I tried out about 4 encoders (not the Fraunhoffer) to do comparisons; I plumped for the Xing due to the results, the cost ($30), and the speed with which it works. So it's primarily money and time to experiment. If you were looking (as I guess you will be, Doug ;^) for purely top quality and have no cost or space considerations, then the Fraunhoffer would be the one. It is $$expensive$$. Bear in mind that RealNetworks (nee Xing) are supposedly trying to licence the FH codec for Audio Catalyst, then for the current cost of AC I would say buy that now to experiment, grab WinAMP for your PC to try things out, and then when you have the empeg, decide what to do then.