I have been shopping for a portable mp3 players for some time now. I gave up on Creative Hard Disk based players because they have serious driver issues and Apple iPod because of the price.

I like the iRiver players very much and would probably purchase either the 40gb HD player or 1gb flash player.

I noticed that all Creative players have > 98db Signal to Noise Ratio, which all other players have anywhere between 30-50db SN ratio.

[EDIT: SNR ratio of the FM tuners are 30-50db the SNR ratio of the mp3 players are 70-90db]

I have not found any mp3 players that even comes close to matching Creative's sound quality.

How bad is a 50 SN ratio? How important should the SN ratio be in my list of priorities when choosing a player?

I should mention that I rip my own CD's, so I don't need any (*)lame support for any online music scams (like napster).

(*)lame as in sucks, not the encoder.

A bad cassette deck with the Dolby off has a better SNR than 30-50db.

Are u sure?

Ok my bad, the SNR ratio of the FM tuners are 30-50db the SNR ratio of the mp3 players are 70-90db.

Some players try to hide the SNR ratio and does not list it in the specifications, so I have to hunt for them.

Well, you need to be careful when comparing different S/N ratios because the way that number is tested can make a big difference.

First off make sure they are both "A-weighted" which adjusts for your ear's varying sensitivity vs. frequency to get closer to what the "perceived S/N ratio is." In general, an A-weighted S/N ratio will be higher than non-weighted.

Another consideration is whether the audio signal is sent through a low pass filter before measuring the S/N ratio. Some engineers LPF the signal below 20kHz, some 15kHz, etc. to make the mesurement - the more aggressive the filer the better the S/N will be as you have remove ultrasonic/high frequency noise.

The difficulty will be that many companies do not specify either of these or other test parameters, so one comapny's 98dB may be another's 89dB. Hard to say without the data - you're on your own.

Of course a higher S/N ratio should sound better all things being equal, which they rarely are, and assuming that the ratio is not already extremely high (like 120dB vs. 110dB should be a wash).

You may want to consider what environment you will be listening to your music in and how loud you are going to play it, which will both affect how audible noise is. If you listen in an anechoic chamber at high volume then you'll hear the noise floor. If you listen on the subway you probably won't.

50dB is pretty noisy, cassette w/Dolby B should be in the mid 60's, CD is theoretically about 96dB. In my experience anything above ~ 85dB is probably pretty acceptable for portable applications. I don't think FM ever gets above about 70dB in stereo which should sound okay for casual listening.

There are a lot of other factors that can have a big impact on sound quality - low frequency response is a big one for me. Often low cost electronics skimp on the output coupling capacitors and roll the bass off too early and sound thin. Of course a poor high frequency response can sound dull. But for all measurements you could have concerns that the tests are done differently and apples to apples comparisons are difficult.

I would say S/N ratio is one data point. There are others to consider as well (price, size, battery life, display/ergonomics, PC interface, other audio specs, quality, etc.)

Edit: S/N ratio examples added

The Rio Karma has outstanding sound quality and is quite likely the best all-around 20gig player:

Thank You! That helps a lot.

Anybody has the details about the measurement of the A-weight SNR ?
Can be executed on PC ?

For an input audio of 16 bits, a SNR of 98 dB seemed to be an overkill.. Actually, you can't hear the difference anyway.. since the dyanamic range of 16 bits audio is just 96 dB!

I've heard good things about the Neuros for audio quality. I'm getting one and I'll be measuring the SNR, etc. as soon as it arrives.

As others have pointed out, 96 dB is the most that could be expected from a 16-bit audio source (theoretically perfect CD) and I have yet to see a 16-bit system that really gives you 96 dB SNR. So if they claim 98 dB SNR I'd be curious to know exactly what their test procedure was.

There is a very easy to use audio test program called RightMark Audio Analyser (RMAA). You can use it to test sound cards, and also MP3 players (if you have a good enough soundcard to measure the output without distorting it). RMAA can give you plots of SNR using the "A" weighting curve, and it is a free download:

http://audio.rightmark.org/

In my opinion, anything above 90 dB SNR is excellent as far as that parameter goes, but there are others: frequency response, distortion, and transient response, to name three. (Transient response is mostly a problem with speakers, and headphones- usually the electronics is fine there.)

As it was an A-weighted measurement it could easily have a figure better than 96dB while still having a noise floor above (i.e. worse than) the theoretical limit of CD.

Cheers,
David.

Yes it is possible- according to RMAA the max achievable SNR with 16 bits and (A) weighting is just under 98 dB. So A weighting does gives you an advantage over unweighted, but it is less than 2 dB worth. In the actual hardware I've tested, the only systems that exceeded 96 dB(A) SNR were using 24 bit, 96 kHz or better DACs. I'm not that familiar with MP3 players overall. Are they using that kind of DAC now?

Be careful (and skeptical) of company claims regarding SNR. For example, Creative claims "98dB Audiophile Quality!!," yet I was able to easily discern cracking and rumbling interference from the HDD motor on a Nomad Zen every time the damn thing spun up to grab more music (which was quite often with my ~240kbps files). In short, Caveat Emptor.