By James Kim and Michael Degnan (April 25, 2006)
Last year, a group of CNET editors listened to a lineup of high-profile, high-capacity MP3 players and rated them based on sound quality
. Purely a subjective (though blind) comparison, the experiment involved listening to high-quality MP3s in a variety of genres on players set at factory settings. Sony's NW-HD5 consistently came out on top, with the Cowon iAudio X5 and the iRiver H320 close behind. Apple's iPod Photo, though not far from the Sony in terms of sound quality, usually placed last.
The winner of our human sound test: the Sony Network Walkman NW-HD5.
What sounds good to human ears may not necessarily sound good to a hard-core audio analyzer, which is able to detect noise, distortion, and other signals that are not true to the original audio source material. And thanks to the folks at Audio Precision
, we were able to borrow an ATS-2A
in order to test a group of MP3 players in an objective manner.
For our initial test series, we pitted 13 popular MP3 players against one another. Rather than testing each player at maximum volume (which differs widely among brands and models but also offers the best performance of an individual player), we analyzed a test tone at a consistent volume of -9.5dBV to 11.0dBV--on the loud side of typical listening levels--so that we could better compare players. We will publish results from the former methodology as well as results from additional players in the future. How we tested
Quantitative audio performance tests were performed with an Audio Precision ATS-2A audio analyzer
using control software and QuickTest 1.01 for personal media players. We connected each player to the instrument in parallel with a 30-ohm load, to simulate earbud-style headphones. All devices were tested at an output level for -9.5dBV to 11.0dBV (approximately 300 mVrms). For each device, we report frequency response, distortion, and noise in the presence of signal. The QuickTest application tests a device by playing back a complex stimulus signal that contains component tones at approximately 20Hz, 1kHz, 10kHz, and 20kHz. Frequency response is a measurement of the deviation from flat of the response across the audio band; lower numbers are better. Distortion is a measurement of the total distortion and noise relative to the output level; higher absolute numbers are better. Noise in the presence of signal is a measurement of the noise relative to the output level; higher absolute numbers are better. Conclusion
All in all, the results at the high end aren't shocking, with the Archos Gmini 402 Camcorder, the Samsung YP-Z5, and the Apple iPod Shuffle performing the best overall. A positive surprise includes the Sony PSP, while the iRiver models ranked the lowest, despite our subjective opinions of the T10 and U10 being much higher. Audio quality starts with the player's audio processor (DSP) and the audio file. Apple's iPod Nano famously use a PortalPlayer chip, the Shuffle and the Samsung utilize a Sigmatel chip, and Cowon's iAudio U3 and SanDisk's c100 series use processors from Telechips. But the way a player is constructed (in other words, if there is LCD or hard drive electrical interference) will affect overall sound as well. Stay tuned as we continue to update these charts with MP3 players, PVPs, cell phones, and other audio playback devices.
How smooth is the sound of your player?