MP3 Encoding -VS- Other Formats
 

I was having trouble posting a reply in another thread, and seeing how this is a different subject than i originally started over there, it is fitting to start an entirely new thread.

I'm sure Lord Of The Rings will find this thread, he/she seems fairly active in these forums.

The discussion regards MP3 encoding to be inferior to lossless formats, and i am starting to agree. Last night i was playing with some Windows Media Player 10 features such as Quiet Mode, Auto Volume Leveling, and most importantly, SRS WOW effects.

I won't go into details about each one, as i'm sure they explain themselves especially to anyone familiar with WMP.

First, know that I record music as a small-time hobby and a big-time dream. This affords me the rare glimpse into behind the scenes problems with recording, producing, and mastering music to perfection. It is not an easy thing to achieve, hence the high dollar value nature of the business.

Usually I mic a person wearing headphones, and depending on certain settings and situations the mic will pick up the headphones and other ambient room noise like the computer and other equipment (my fault for not having a soundproof room). To eliminate these noises I run Noise Reduction over the track. If it is not done perfectly, there is a noise called "burble" introduced, which is usually inaudible except at louder volumes. The point is, there is a strange noise there that interferes. For an example on this sound, try listening to Launch Cast Radio from Yahoo! on a free account, which is by default broadcast in "medium quality." If you are listening to a track that you have heard on the radio or off the CD then you will know the extra noise that I'm talking about.

Back to last nights tweeking experiment. SRS is probably the major key to bringing out the "burble" noise in many of 5000 MP3/WMA tracks I have ripped onto my computer at 128bit compression. So I feel the simple fact is 128bit compression is not equivelant to CD quality. If it were, then why would WMP have a mathematically lossless format of 470 - 940kBps??

I am at this very moment ripping a CD into Lossless 470-940kBps, and it is taking very much longer, first. Second, this format has a file size of 27.8Megs for 4:49 minutes as opposed to 4.44Megs at 128kBps and 4:50 minutes.

The benefits for having the lesser quality are first the medium users wish to play their tracks, that depending on their usage poor quality may go unnoticed. The second benefit is sharing, since smaller files transfer faster. But, since patience is a virtue... well you know what i'm going to say.

The other thing to worry about is harddrive space. I have 250 Gigs of storage, and with everything in 128bit i have hardly dented my quota. I could, I think, store 1 million tracks of music on this disk at 128bit, and that seems beyond my ability (:mad:). On the other hand, not everyone has such massive storage space, or a DVD drive to back things up on. 5000 tracks fit on 3 DVDs (4 gig capacity), or 15+ CDs (700 meg capacity), take your pick.

If i were not the type concerned with other people, i would gladly rip CDs at the highest and best quality possible, and to hell with you and your patience. Yet I find convenience in lower quality and smaller files.

But as another experiement I think I will re-rip some CD's at the greater file size/quality and see how it goes with people downloading them. Don't most people burn CD's immediately after they get all the tracks then delete them off their computer? If that be the case then there should be a swelling of uploads on my part...

Peace.:D

I totally agree that that was part of the argument that mp2 & mp3 (particularly in the early days when less than 128 kbps was available to encode), were efficient at transfering music at small sizes & at high speeds across the net. But nowadays with the ability to encode mp3's at up to 320 or aac/m4a/mp4 at same but smaller sizes AND with much larger HDD sizes available AND much faster net transfer speeds that people (the ones who do care & notice the difference) are very definitely looking at higher bit rates to downld. Personally I encode 80% at 192 kbps & the other at 320. I went thru the experiences of encoding using lame & Og Vorbis & others using various apps & encoders, at suggested low bit-rates. But when I listened to particular (& I specify particular) pieces of music which had an extremely wide dynamic frequency range & stereo separation, I thought there's something missing here, & then listened back to the originals on CD & realised exactly what type of compromising was being made. It does very much depend on the particular music/song you listen to be able to pick up exactly how much of a difference it makes.

The Trade Off

Decisions regarding encoding and compression involve balancing 3 factors:

1. Audio fidelity/quality
2. File size [affects hard drive space and Internet transfer time]
3. Time [to encode/compress]

Format tests: (take these tests with ... very informal tests ... anyway top-down in quality of performance.
1. # MPC: 4.31 (143kbps) (mp3pro which has best results at lower bit-rates.)
2. # AAC: 4.12 (132kbps)
3. # Ogg: 3.48 (133kbps)
4. # Lame: 3.36 (128kbps)
5. # WMA: 3.23 (132kbps) - some complain of squishy
artifacts below 160kbps
6. # Xing: 2.41 (128kbps)

Related post: Music Quality (popular demand) Your turn: http://techhelpers.net/e4u/drink/trink22.gif http://techhelpers.net/e4u/drink/trink21.gif

By the way miking is an art in itself. It's like or should say is an audio engineering skill or qualification. People in years past have learnt it not thru school but by learning it on the job. But then every new model that comes out is just that much different, despite its aims at times. It's too broad a topic (even broader than mp3's) to discuss here. But I guess the trick is to try to learn to use the equipment you have available in the most skillful & effective means possible to gain the end result. Sounds obvious but isn't in practice. eg: you can artificially set up directional microphones. Else you can artificially set up bi-directional mic. But using multiple mics can in itself produce major issues such as phasing b/w the mics & being out of phase, & also dead spots, etc. (Angle/direction & type for freq. capture.) Keeping in mind that sounds can cancel themselves out! blablabla I think you get the drift. hahaha

If you were to encode to lossy mid-range bitrate this would be a good way to go:

Use mpc encoder (musepack)
Based on mp2

At
--quality 6 --ms 15

Achieves around 200k average and is the most transperant codec to me and others!