Multiformat Listening Test @ 48 kbps - OPEN


I would like to announce the launch of my public, multiformat listening test @ 48 kbps.

The encoders featured are:[ul]
[li]iTunes AAC (96 kbps High Anchor)
[/li][li]Ogg Vorbis AoTuV 5 Beta
[/li][li]WMA Professional 10
[/li][li]Nero HE-AAC May 26 2006
[/li][li]WMA Standard 9.2
[/li][li]iTunes AAC (48 kbps Low Anchor)
[/li][/ul]For further information on how to participate, please visit the test announcement page:

The test is scheduled to end on December 8th, 2006.

For those using BitTorrent, statistics are available here:
Please use BT whenever possible.

Thank you!

Sebastian Mares

What’s this now?



Looking forward to seeing the results. I’m rooting for Vorbis. I don’t expect it to win, but at the same time I think it will fare pretty well.

The much awaited results of the Public, Multiformat Listening Test @ 48 kbps are ready.

Here is the results page:

Nero is first, followed by Vorbis and WMA Pro. which are tied on second place, WMA Standard is third and loses.

I think this test shows that with modern encoders, the quality at 48 kbps is acceptable and should be good enough for Internet streaming or portable use with cell phones for example. It’s also interesting to see that WMA Professional perfomed quite well although it was the only contender that used CBR.

…I hate to say it but those results wouldn’t lead me to that conclusion. They are subjunctive results, not absolute. You’re comparing audio formats at low bitrate to others at low bitrate. They are only being compared to each other and cannot prove that “the quality at 48 kbps is acceptable” as there was no high bitrate option for people to choose. At the end of the day, music encoded at that bitrate can’t compare to music at 128 or 192kbps or above, and such low bitrates as 48 were really only intended for speech recordings.

I can agree that these files are probable “good enough for Internet streaming or portable use with cell phones” though, as you don’t expect quality from those things (at least until we get even faster internet connections which can stream at high bitrates with no problems).

Of course there was - the high anchor iTunes LC-AAC at 96 kbps. There was also a low anchor: iTunes LC-AAC at 48 kbps. The anchors are not seen in the zoomed plot.

Did you take the test? Also, as my last listening test at 128 kbps demonstrated, bitrates such as 320 kbps, 256 kbps or even 192 kbps are overkill [B]most of the time[/B] (not always - there are some killer samples that demand for higher bitrate but these are very rare) because all contenders ended being tied on first place with ratings around 4.5 from 5 points.

There’s no hiding the fact that when you start going beneath 192, frequencies are cut that are in your audible hearing range. It does sound different and no test will change that.

My last 128 kbps test showed the contradictory. Anyways, I will stop arguing because this won’t lead to anything. An audiophille will also never admit that his 50.000 bucks RCA cable sounds just as good as his neighbor’s 25.000 bucks RCA cable. :smiley:

Oh, I’m not one of those people (who fuss about paying stupid amounts for a wire) :stuck_out_tongue: and I also don’t want to argue. I’m more fussed on the encoding being done the way I like it. It’s just a physics thing, the human ear can distinguish some of the frequencies lost below 192, but when you start hitting 20khz+ (or a little less) then it’s out of our hearing range.

Erm, somehow this steered away from “the quality at 48 kbps is acceptable” to “the human ear can distinguish some of the frequencies lost below 192.”

I think the results do in fact show that generally speaking, especially for Nero HE-AAC, the quality at 48 kbps is indeed acceptable. The anchor scored in the same range (4.69) that the participants in Sebastian’s 128 kbps test did, and those codecs were pretty much transparent to most listeners.

HE-AAC’s score of 3.64 is definitely lower, but respectable. It would be interesting to compare it against something like Atrac3 at 130 kbps. I bet it would be close!


could you tell me the filesize of each files? (especially Ogg and Nero’s HE-AAC)… thank you :slight_smile:

Right, I’m not buying this, when I encode at such low bitrates the trebles are all gone from the music, and the music sounds artefact-y in many places (more complex zones of the music). There’s no denying it, and I’m listening to this with my own ears, which is proof enough for me. Is there really any point in crippling our music with such crappy bitrates when we have so much potential in storage capacity?

Hear, hear!!! Now, way off topic:
Many of us “audiophiles” took Pete Townshend’s advice years ago and wore earplugs whilst standing next to jets, operating chainsaws, or attending Who concerts. Thus, we can hear well into the 20kHz range. My last hearing test, I was still raising my hand at 22kHz and I am 42 years old. Personally, I’ve never listened to a cassette tape (max. frequency ~17kHz) that I didn’t think sounded like the band was recorded under a carpet. Furthermore, recorded frequencies beyond human hearing, yet reproducible with even moderately high end equipment, can resonate with frequencies that are within hearing range and make them sound “louder” to the human ear. Finally, digital sampling at rates lower than 2X the maximum analog frequency in the recording can cause aliasing that cannot be filtered, usually leading to the high pitched ringing or squeaking that I begin to hear at those sub-192 mp3 rates. FTR, that’s why CD’s are sampled at 44kHz.

These bitrates are only useful in bitrate / capacity sensitive situations, and where the choice of format is not already pre-ordained by the hardware.

48k is a little too high to be reliably streamed by “56k” modem.

Another point that online broadcasters generally try to go for, is how it compares to MP3 streaming.

And finally…

DAB - Digital Audio Broadcast

The version 1 system in use athe moment in the UK, uses MP2 - a codec which is far less than optimum, at bitrates that are also inadequate, from BBC’s prime stereo quality at 192k maximum (drops to 160k to make room for another voice channel), down to 128k or less, even for stereo.

Version 2 in a little publicised announcement (since they would rather not see a slump in sales of “to be obsolete” DAB V1 systems), will use AACplus, though probably to increase capacity a lot, and quality only slightly.

It’s probable that AAC+ and MP2 would actually run in parallel, with any new stations operating AAC+ and MP2 being progressively phased out.

Codecs with lowered frequency cutoff are obviously not aimed at users like yourself with [I]‘golden ears’[/I], but for average persons who cannot detect so high :stuck_out_tongue: . Even LAME MP3 with most preset settings has cutoff below 20khz.

Sebastian Mares:[/B]

How come CTI AAC codec was not tested? It is about equal to Nero AAC in most situations.