Lossy audio formats reduce the high frequency tones because high pitch requires much more data to preserve than low pitch tones. Humans can’t hear the tones and speakers to do not reproduce tones. Humans can’t hear above 20-22 kHz and adults normally can’t hear above 16 kHz. You really can’t even hear high pitch tones near your hearing threshold unless they are real loud (Fletcher and Munson). You will not discern these tones in your music even if they were present and your ear can physically hear it. Contains hearing limitations and Fletcher and Munson equal-loudness contours http://en.wikipedia.org/wiki/Psychoacoustics 20,000 Hz contains 1,000 times more data than 20 Hz. To get a faithful reproduction the sample rate must be much higher to the total cost in bits is much more than 1,000 times. It is also believed by many and is supported in real life that ultrasonics are more damaging at the same db level than audible sound. If you are a speaker maker, you need to consider the liability for making a speaker that produces ultrasonics that no one can hear but will damage your hearing more effectively than if you left them out. It is foolish to use most of your band width for tones that will not be reproduced by any speaker or ear bud. The CBR mp3 uses that principle in making a lossy reproduction of the sound. It cuts off the highs to maintain a particular bit rate. For this reason there is not an absolute high end cut off but it normally runs about 20-22 kHz for a 320 CBR. Even a baby would not miss a note. Lossless - That is really a misnomer you always lose data going from an infinite stream to a digital stream. For this discussion you can just think of it as a 1200 - 1300 CBR. Persons I fondly refer to as analog junkies can point out vinyl holds more data than a CD, so it is better. They are correct about in holding more data. Better, I don’t think so unless you feel the pops and hiss need to be part of the listening experience. What good is data you will not use? Psychoacoustic compression VBRs, AAC, WMA and many more formats - See the Psychoacoustic reference for detailed info. These formats use the most advanced kind of compression removes tones your brain will not use even if it receives them from the ear. Formats using this technique have a better quality to file size ratio than formats that do not. I feel compelled to put my opinion forth since I think it might be helpful to some. I believe the VBR format is the best controlled format. It is the only audio format that the creator has the best control of the quality of the music. They do completely relinquish the control of the bit rate (Variable Bit Rate). Psychoacoustic compressionbeing complex, is the most prone to artifacts. An artifact is noise created by the encoder because it wasn’t built right. You need to be more selective in your choice of encoders if you want to avoid artifacts. LAME is the undisputed king in this realm. It was thoroughly debugged years ago. A pitiful few isolated finds were reported but no one as reported any artifacts in years. The next best still has artifacts. The last time I checked, Apple, M$ and Real Media formats have long lists of complaints. LAME quality settings http://wiki.hydrogenaudio.org/index.php?title=LAME#Technical_information I think I can hear tones you can't - What is the cut off frequency for me? There is a hearing test Mosquito tones and the student repeller device that resulted in Mosquito Ring Tones Mosquito Ring Tone Story http://www.freemosquitoringtones.org/ Hearing test Ultrasonic tones damage ears even though the tones are below the loudness danger threshold. http://findarticles.com/p/articles/mi_hb6368/is_4_76/ai_n28968663/]Ultrasonic tones damage ears[/url] I just cleaned this up a bit since I am going to link it.