Notes by ff123
4/3/01: Redid the listening tests of Lame using 3.88 beta. The sound of --nspsytune has not changed appreciably since I last listened to it with this sample.
3/20/01: Using Grado SR325 headphones. Fanless computer operation. Decoding mp3 with Fraunhofer decoder within Winamp 2.73. Using WMA decoder from Winamp 2.5c to decode WMA files to WAV.
This is the first 10 seconds from the Eagles's "Peaceful, Easy Feeling." You can download the clip (peaceful.flac) from my Audio Samples Page. The file has been losslessly compressed using the FLAC format.
I don't know what makes these Eagles clips so hard to encode properly (it's just guitar strums), but all the mp3's at 128 kbit/s sound pretty crummy. MP3Enc really sucks on this sample -- far worse than Xing. If I had to choose a best mp3, it would probably be FastEnc CBR or VBR 50% over Lame with --nspsytune. That's because the noise caused by --nspsytune is really starting to irritate me. I feel the same sort of unease about the lack of clarity that one feels when listening to someone talk who needs to clear his throat. Ogg Vorbis beta 4 sounds pretty good to me, except for a noticeable smearing of percussion. WMA sounds pretty good to me as well, but looking at the spectral graph, I have a niggling concern that others who can hear high frequencies well may be bothered by ringing -- I don't know for sure; it's just a guess.
On WMA:
Addendum: bAdDuDeX writes (original post on the VQF forum): I listened to the sample and didn't hear any ringing. I've never heard ringing with any codecs except Blade (ISO), LAME, and FastEnc. However, the WMA sample did sound messed up. It had some added noise in the background and a somewhat slight metallic sound. It basically sounded weird overall though.
Nitemare writes on the r3mix forum (original post here): I encoded a randomly selected song at 192kbps. Throughout the entire song, I heard high-frequency background noise. It sounded like wind-chimes on a tape being fast-forwarded. It was really weird.
Added test of Psytel AAC and Liquifier Pro 5 AAC
| Encoder | bitrate | settings | ABX results | Comment |
| Xing in AudioCatalyst 2.1 | 128 | no high freq | didn't bother | obvious flanging |
| FhG mp3enc31 | 128 | -qual 9 -bw 16500 | didn't bother | worst by a large margin! bad chirp at the very beginning; ugly noises and dropouts all over the place |
| FhG 1.2 build 63i external codec (Radium) | 128 | didn't bother | flanging; chirp at beginning; not as bad as Xing, and better than mp3enc31 | |
| Lame 3.88b | 128 | -h --nspsytune --athtype 2 --lowpass 16 | didn't bother | characteristic --nspsytune low-frequency noise |
| Lame 3.88b | 129 (average) | -h --abr 134 --nspsytune --athtype 2 --lowpass 16 -b112 | 16 of 16 | same as above |
| FhG "FastEnc" codec (EasyMP3 1.1.0) | 128 | "Faster" | didn't bother | chirp on first guitar strum; flanging |
| FhG "FastEnc" codec VBR 50% (EasyMP3 1.1.0) | 132 (average) | Using CBR 128 to set max bandwidth | didn't bother | same as above |
| Ogg Vorbis beta 4 | 132 (average) | 16 of 16 | better than any of the mp3's, not as much flanging, although I can hear smearing (pre-echo) of the percussion. | |
| Microsoft WMA (version 8.0.0.0302) | 132 (average) | 14 of 16 | best to my ears; similar to Vorbis in flanging quality, but less pre-echo. | |
| Psytel AAC (version 1.0, April 19, 2001) | 129 | -br 128 -qual 9 -production | didn't bother | chirp on entry to sample, softer chirping later in song (sound of someone sharpening knives in background) |
| Liquifier Pro 5 AAC (streaming 128) | 128 | default equalization | 16 of 16 | hihat decays have a flanging sound to them. Probably second best to my ears. |