why no hiss?

Two Christmas’s ago I got a used iPod mini. This year a story in mp3 format. According to two different utilities, the mp3 encoding is 32kbps. It decodes to  mono, 41000kHz, 16 bit. The upper frequency limit is pretty sharp, from 7200Hz to 7800Hz.

In spite of the limitations, it is quite clear, and easy to understand. However, listening to it via my computer setup -- audiophile 2496, Graham Slee headphone amplifier, Sennheiser HD 600 phones -- there is a strong hiss that I feared might be quite uncomfortable after awhile.

Being spoken audio only, there are plenty of pauses containing only the background noise to analyze. Although that noise level is considerably lower than the speech, it is much higher than the LP or cassette music with which  I am used to working, especially above a few hundred hertz.

However, after loading the mp3 files onto the iPod, I can’t hear the noise at all. This is the mystery I present. What can account for the difference?

I tried the HD 600’s on the iPod, in place of the in-ear phone I use there. I detect no significant difference, I still don’t hear the hiss.

I attached the in-ear phones to the headphone amp and played from the computer. At the same volume setting, the hiss is a little less loud than with the Sennheisers, but not greatly different.

On computer, I tried a fairly sharp FFT low pass filter (FFT size 8960) in 1kHz steps, leaving the volume control untouched. There seemed to be no change in the hiss until 4kHz, when it became definitely quieter. At 3kHz it is practically gone, but the spoken audio is significantly deteriorated, which does not happen when played back on the iPod. Overall level of the noise is down only a couple of dB after the 3kHz filter, as measured with CoolEdit’s Statistics.

I don't know definitively, but there are only two possible choices (or a combination of both) - either downward expansion in the amp, or a very carefully profiled EQ curve. We have in the house an iPod shuffle, and on the only occasion I've ever listened to it, I came to the conclusion that something weird was going on, and along very similar lines to what you are suggesting. Since my desire to listen to something that distorts like that is zero, I dismissed it out of hand, and haven't given it another thought - it's just more Apple-hyped junk as far as I'm concerned.

I don't propose to do this myself, but the method of testing this should be quite straightforward, I would have thought. Known noise samples at different levels on a file, and measure what comes out of the headphone socket - that should establish what's going on.

I’ve thought about recording from the iPod but haven’t worked up the energy for it yet.  When I am working every day, as I have been for awhile now, (often seven days a week) I have much less energy for playing with the computer. 

The soundcard input impedance is quite different than that of headphones. Is there any aspect of a recording from the headphone jack that might be misleading because of that difference? 

I’ve never listened to music on the iPod, it always seems of too little interest to bother preparing and loading the files, when it occurs to me at all, but I have listened to a number of audio books. Mainly these help pass the time when traveling. If there is any distortion, it is insignificant according to my criteria. All that is necessary with speech is that it be easy to understand and comfortable to listen to.

I don’t know if this is of interest to anyone, but since I started it, I will report the end of it.

I generated a CoolEdit file, 16 bit mono @44.1kHz. Five seconds of silence at beginning, end, and between each sample, which samples were: white noise, pink noise, a sine wave sweep tone (20Hz to 20kHz @ -10dB), and brown noise. The noise samples were each intensity 12. All were ten seconds long.

The noise sample file was mp3 encoded in CoolEdit at 320kbps, 20kHz cutoff, 44.1kHz sample rate. Both the original wav and the mp3 versions were loaded onto the iPod.

Recording was from iPod phone jack to an Audiophile 2496 analogue input. I also recorded about one and a half minutes of the file from which the sample of my first post was taken.

After adjusting amplitude of the recording to match the original files (the iPod output is through a volume control), I found no differences I considered significant. More exacting standards might arrive at a different evaluation, but such differences as there are probably cannot be detected in double blind listening tests. Background noise, no doubt from both the iPod and the soundcard, was higher than in the CoolEdit generated file, but at -76.6dB peak, -89.3dB average, also seems insignificant.

Noise in the recorded story file measured and sounded the same as in the on-computer source mp3. I can only conclude that I habitually listen to the iPod at a lower level than I use at the computer, where I am usually concerned with detecting noise or other problems, and that I was unable to notice that this was so when I did my original comparison. I can turn the iPod volume up enough to easily hear the noise; it would become tiring and unpleasant in a fairly short time. A hurricane in a tea cup, or however that complaint goes.