Listening Test Using Male Opera Voice

Notes by ff123

12/15/00: Using Grado SR325 headphones. All fans in computer disabled. MAD 0.12.3 beta 4 decoder

This is a piece of opera with male voice. The sample is available on Panos Stokas's site (as "1400.wav"). Panos compares mp3 to Real G2. I didn't realize until I had completed a first round of testing that Panos had "doctored up" the file with phase and stereo effects. Since I went through the trouble of comparing codecs, though, I decided to keep the page.

TLO supplied me with the mp3 files for SoundJam and iTunes. His page is here.

This is the first time I've had trouble setting the volume level. I notice a weird phenomenon with this sample using certain codecs -- I find it easier to hear artifacts with mp3enc and the Alternate codec if I turn the volume down a bit.

At 128kbs, the Radium hack is definitely the worst, with a watery sound. All codecs except mp3enc and Alternate (and perhaps the Radium hack) have a ringing or noise-like pumping that occurs along with the start of the singing. The original clip has a trace of this as well, but encoding makes it worse.

Addendum 01/01/01: The ringing artifact produced by Lame and FastEnc is something which at least two people with good ears (bAdDuDeX and HansHeijden) finds objectionable. See my page "Ringing Artifacts in Lame."

On the particular sample tested on this page (opera with male voice), bAdDuDeX writes:

Hmm...I quickly tested it with MP3Enc too and I thought MP3Enc sounded better on it... There were some warbling sounds with MP3Enc, but it didn't have ringing (like in LAME) and had cleaner vocals. [bAdDuDeX overrode the default low-pass filter cutoff settings to perform his listening test].

Also, Hans Heijden writes:

A few days ago I did some experimenting in CoolEdit with 1400.wav. The ringing is so strong in the left channel that I think I see it in spectral view. Doesn't look as dramatic as it sounds though.

Return to ff123's home page