Audio Compression Blind Test - test your hearing!

[az]

I'm planning on doing an audio compression blind test with you guys as volunteers.

I'll be ripping some music with EAC to .wav and then encode it to different formats and bitrates (I've thought of 128k, 160k, 192k, 256k mp3, maybe VBR mp3, and some .ogg). I'll then decode these files to .wav and upload them with names like test1.wav, test2.wav, etc. You take your time to listen to them on whatever equipment you want and then post your results, from best to worst sounding.

I've thought about also including MAD vs. Nullsoft's in_mp3, since some of you claim that MAD is clearly superior. A few questions: Does MAD work in Winamp 5? What are the settings for best quality? What bitrates should I use? (I'm not going to do all, since it would get tiring soon)

OK. So, what formats/bitrates would you like to hear? What music should I use as source? NOTE: I have no high quality classical music on CD, and don't have that much contemporary music on CD either, but I'm trying to find some kind of common denominator, so post your music wishes, please. Pearl Jam, Chanson, War of the Worlds? Chris de Brugh, New Model Army, No Doubt, Marillion, Faith No More, Rammstein? (I think Rammstein is mastered with clipping, so I don't think it's a good test)

Hope to see much participation

AZ

[gt40]

I think classical might be the best choice, as much rock has distortion inherent in some of the instruments.

I can dump a .wav file of some classical for you on an ftp if you like.

[az]

Yeah, we could do that. Maybe one classical, one pop/rock.

I think acoustic guitar or piano, clear voice, deep drums and hihats would be a nice test, if recorded very well. If anybody knows/has such a piece, feel free to mail to azrael_katz@gmx.net

AZ

[Wombat]

What encoder were you planning on using? LAME?

You know, the first time I really noticed MP3 low-bitrate distortion was with NIN's <I>The Perfect Drug</I>. The song was really busy, and fixed bitrate didn't have the bandwidth to make those hi-hats sound good.

[az]

I was planning on using LAME with stock settings, but I am open to suggestions. I also have no experience with any recent .ogg encoder, so suggestions and help are appreciated.

AZ

[Jon P. Inghram]

As for bit rates, when using LAME I'd just use the VBR --preset combos: medium, standard, extreme and the CBR 320 kbps insane.

[Jon P. Inghram]

Az, check your PM, forgot that the PM system's notice system's not working right.

Err, well... just checked the thread in the feedback forum and it seems it's fixed, doh.

[Mehen]

You need something with both good highs and good lows.

[spadnos]

I did a test like this with a friend. (he provided the Audiotron )

We hooked the digital optical output of the audiotron directly to my Pioneer Elite VSX-39TX receiver, as well as my Pioneer DVD player, attached the same way. I have excellent speakers (Mirage K760).

We only tried with one song, "Behind Blue Eyes", by The Who. We then compared the original CD with the 3 LAME alt-presets (standard, extreme, and insane). We checked the difference in quality by playing the CD and the mp3 simultaneously, and switching back and forth between them at the receiver.

Since we encoded from a digital file (ripped to WAV with EAC, then encoded as bits), and used the optical outputs, the sound card quality should have had no effect (it's an SBLive).

We could pretty easily detect the difference between the "standard" encoding and the CD. We had to listen pretty carefully to tell the difference between the extreme or insane encodings, which both sounded about the same to us.

(talking about the "extreme" encoding level here) The main thing that happened to the MP3s is that the low end sounded "muddy", and the high end lost a bit of punch. I don't think this would be detectable without a pretty high end stereo. I can't tell the difference on my Harman Kardon computer speakers.

Just my EUR 0.02. I'd love to perform the same test with OGG, so I'll be interested in seeing the results of your experiment (and I'll even see if I can borrow the Audiotron again, so I can participate).

- Steve

[Jon P. Inghram]

Since we encoded from a digital file (ripped to WAV with EAC, then encoded as bits), and used the optical outputs, the sound card quality should have had no effect (it's an SBLive).

[audiophile prick mode: ON]
LMAO, you're such a noob! If you didn't spend at least $500 on fiber optic cable and an expensive "jitter corrector" you couldn't possibly get good sound!
[audiophile prick mode: OFF]



When using a Live! card beware of the stupid AC97 sampling rate conversion issues... I've never understood what the point was behind that crap.

[GuchiGuh]

am in.

Like to see: 128kbps, 192kbps, 320kbps (Mp3 CBR), VBR, Ogg (i've never heard ogg so I'd be interested)

Any music would be good, as long as they tests out much of the audioable spectrum!

I'll help you host the mp3s and make a webpage as well if you need some help Az. Just let me know~

[Brian R.]

How about a sine wave at 440 Hz or nearby. You could actually visuallize it on an oscilloscope and see differences.

[KvHagedorn]

If you want a good source, use YoYo Ma's newer recording of the Bach suites for cello. Original is a 20bit recording which sounds incredible.

[Jammrock]

I'll test I've got a pretty good audio setup and a decent set of ears. You should also throw in Shibatch mpg123 as the Winamp MP3 decoder, as it has even better sound than MAD.

Jammrock

[Jon P. Inghram]

I'm using the Shibatch mpg123 with the DirectSound 2.2.6 SSRC output resampler, sounds pretty nice.