find out all about mp3 encoding quality

June 21, 2012 at 04:34 PM · To find out the truth of mp3 encoding quality compared to original, non-compressed sound files I have prepared a comprehensive and objective test for you here: http://bit.ly/Lm0tGC

Please share the link as much as possible via facebook, twitter and alike to broaden the base for the poll.

FMF

Replies (56)

June 22, 2012 at 05:36 PM · MP3 is rubbish, so don't bother with it.

June 22, 2012 at 06:39 PM · Great, then tell us, which of the 12 is NOT mp3! ;-)

FMF

June 22, 2012 at 07:21 PM · "(...) il fit ce que je voudrais faire

Si j'avais quatre dromedaires."

Thank you for putting in the time, the energy and the initiative. I'll see if I can solve this puzzle.

edit, after listening to the first five minutes of all takes: The differences are often subtle. It took me a few takes to recognize the English horn near the beginning. So I made that Aha take my favorite.

June 22, 2012 at 08:10 PM · I can't be bothered. In any case the first one was such a large file that I would be all day downloading it, and I have a fast computer and internet connection.

June 22, 2012 at 08:12 PM · Lyndon, these are .wav files (CD quality) so that we can (maybe, at my age!) distinguish between the different compressions.

My computer sound-card handles .wav files perfectly.

June 22, 2012 at 08:27 PM · I assumed that they were .wav files to conceal the information about what type of mp3 they were i.e. if they were there as mp3 all one would need to do would be to open them up in vlc player and look at the statistical info, which would defeat the purpose of doing it by listening.

June 22, 2012 at 08:31 PM · I made a CD containing the first five minutes of each take. The only way that involves the computer is as a data copying machine. I believe it can be trusted in that capacity. I listened to the CD on the living room stereo. And still it was difficult! Must be my old ears.

June 22, 2012 at 08:54 PM · Nigel, I suspect the different sizes of the files have some bearing on the quality, although I suppose 96kHz at 128kB/s willhave the same total as 48kHz at 256? Anyway, I promise not to cheat! But the dowloading is taking ages..

June 22, 2012 at 09:08 PM · Anyway: after June 30 I shall publish the encoding parameters behind each of the files. One is actually the original right out of the recording studio. No compression was involved at any time there. All others (11) have been "compressed" exactly once.

FMF

June 22, 2012 at 09:23 PM · Am I allowed to boost above 10kHz on playback to compensate for my ageing ears? At 63, I have nothing, but nothing, above 13kHz. I shall use headphones. Of course it won't be just the highs, there will be a question of tranparecy vs muddle.

June 22, 2012 at 09:27 PM · You may play back with volume and equalizer settings you prefer. Just make sure to use the same for all 12 samples.

FMF

June 22, 2012 at 10:48 PM · Adrian, I can trump you on that! I hear nothing above 10KHz, which, to put it in perspective, is about an octave and half above the top A on the piano, so it seems pointless to me to try this mp3 comparison exercise. It also follows from my top-end hearing loss that when recording live for myself I do not need to use sampling rates above 22KHz (the highest recordable frequency is half the sampling rate).

June 22, 2012 at 11:21 PM · Meanwhile for a change some live recording produced about three weeks ago:

Julia Fischer performs Mendelssohn's Violin concerto in E minor

Details when clicking on the link above.

FMF

June 23, 2012 at 06:54 AM · Trevor, I was boasting! I need a hell of boost to hear 12-13 kHz. However, the apparantly high sample rates can have an impact on the clarity of the sounds which we (still) hear, just as a high definition photo looks sharper, even if I can't see the pixels (even with my increasingly powerful reading glasses!)

I shall be listening for the purity of the solo sounds and the transparency of the ensembles. The mp3 compression doesn't just remove high or low frequencies, it tries to remove sounds it "thinks" will be masked by others: a real mash-up.

However, remember the dolby-ised cassettes...

June 23, 2012 at 07:00 AM · My ears are really trying.... I do hear a difference, if my ears is not tricking me, but which one is better? Hmm...

But what I do know is that it is JF playing Respighi. Do I get extra points?

June 23, 2012 at 07:07 AM · I simply record at 24 bits either 44.1 kHz or 48 kHz (.wav files) and then I know I have plenty of headroom with a very low noise floor and pretty clean undistorted recordings. (Using good mics and pre-amp).

And I use large hard drives so there are no space worries. Why bother with mp3? (I suppose people will want to send samples by email or downloads so the small file size would be the only advantage, but even here how long does it take to download a *reasonable* sized ,wav file? If the files are large then give or send a CD or a DVD)

June 23, 2012 at 07:48 AM · I do everything as Peter just wrote. I only use mp3 demos for mail attatchments, or to put onto my students' eye-pods (no advertising) or eye-fones (ditto).

June 23, 2012 at 12:32 PM · Adrian - I'm pleased I am a really strong influence on you. You can't really go wrong following my advice !! (wink)

June 23, 2012 at 12:45 PM · I've downloaded the files (and doubled my electricity bill!), but after several tries, samples F and L are empty. They play OK on the site, though, with some stuttering. Perhaps they are huge files? Aha!

June 23, 2012 at 01:24 PM · None of the files are empty. They are all of equal size: 153276460 bytes (which corresponds to 147 MB), and they are all wav files.

One is the original wav file directly from studio, the other wav file are made (clearly lossless) from eleven different mp3 versions of this original wav.

FMF

June 23, 2012 at 01:58 PM · All mine downloaded and played with no problems. With my ears I couldn't tell which was the original studio recording from among some which seemed to be the best, so I guessed. No, I'm not going to tell!

June 24, 2012 at 05:14 AM · My intention was not to make someone "tell". This test page carries an anonymous poll (the blue field on the right side).

Thanks for participating and keep sharing, please.

FMF

June 24, 2012 at 09:45 AM · I'm still getting 0 MB in F and L! The other files download OK. Anyone else?

June 24, 2012 at 10:04 AM · Here are the download adresses for:

F: http://candidate.dyndns.org:8080/~mifi/final/8e0ad09bf8b1807538d5fb66fb83d2dd.wav

and

L: http://candidate.dyndns.org:8080/~mifi/final/fc13918f4d532b96b360f1b21a823c8a.wav

FMF

June 24, 2012 at 10:57 AM · I've just downloaded one - F

Are we meant to comment on the performance? Probably not.

No idea what F is (wav original or mp3) but the sound was not very good - fuzzy and unfocussed.

I haven't voted because I haven't heard the others apart from F.

June 24, 2012 at 11:11 AM · After all this comparing, I pushed the wrong button! But still I know which version I liked best - just didn't vote for it.

June 24, 2012 at 01:10 PM · Gottit! Or rather, Gottem!

Now I can get down to work (with some trepidation!)

June 24, 2012 at 10:34 PM · I have been thinking about the parameters of this voting test. Basically, its results depend on the hearing ability of the individual and the quality of their playback equipment. I think this skews the results in favour of people with good hearing (mostly younger and with insignificant frequency loss), and possessing better than average speakers/headphones together with appropriately matching amps.

I have devised a simple test to illustrate on a computer screen the visual differences between recordings made by the various mp3 formats and an original wav recording made to the standard CD specification (stereo, 16-bit, 44100Hz sample rate). The point about my test is that there are no emotional or other distractions present caused by real music or ambience.

The equipment used was a Windows XP computer, with a 1280 x 1024 monitor. The audio editor used was Cool Edit 2000 (CE2000) by Syntrillium, and other software used was iTunes.

A 10-second white noise wav file in CD format was prepared using CE2000. White noise has a spectral frequency in which there are equal proportions of all frequencies. Because human hearing is more sensitive to high frequencies, white noise sounds "hissy". CE2000 generates white noise by choosing random values for each sample.

iTunes was used to encode the white noise wav file as 16 files comprising the following available mp3 bit rates:

16, 24, 32, 40, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320.

Variable bit rate encoding was not used, as being inappropriate for a randomly generated sound file.

The mp3 files were loaded back into CE2000 and re-encoded as CD format wav files.

These CD format files were appended to each other in ascending order of bit rate using CE2000 so as to make one big file, the final file in which was the original unmodified wav file. A 1-second gap was made in the big file every 10 seconds to enable the mp3 sections of the big file to be clearly distinguishable on screen.

The mp3 sections of the file were examined using CE2000's spectral view. The highest frequencies for the mp3 bit rates were as follows:

16 - 48: 4.8 - 5.2 KHz

56 - 64: 9.6 - 10 KHz

80 - 112: 15 - 16 KHz, with a few frequencies up to about 19 KHz

128: 19 KHz, with a few frequencies up to about 20 KHz

160: 20.5 KHz, with a few frequencies up to about 21 KHz

192 - 320: all topped out at 22 KHz, as did the original unmodified file.

Conclusions

On frequency range alone, this test indicates that mp3 bit rates up to and including 112 kbps should be easily distinguishable from the original, and it may be considered that listeners with average hearing should be able to distinguish 112 - 160 kbps from the original with very little trouble.

Interestingly, bit rates from 192 - 320 have the same top frequency as the original, and this accords with my difficulty in distinguishing between them by ear. What in fact distinguishes them from each other and the original visually on screen are subtle differences in the patterns of the pixels, which in the aural world would correspond to subtle differences in the dynamics of frequencies and hence tone colours. I therefore believe it would take a very good and experienced ear, coupled with high quality audio playback, to distinguish mp3 bit rates 192 - 320 (especially 256 and 320) from a CD quality original.

June 25, 2012 at 06:57 AM · Trevor, I think the point of the exercise is to test our powers of discrimination through the "filters" of our own hearing and equipment. Practicing musicians are often satisfied with less-than-perfect reproduction - except for their own instrument!

I use white noise often when judging headphones etc., but mp3 compression does more than limit the high frequencies, it messes around with them all. Even a solo violin can sound not just muffled but "dirty", and complex textures become confused rather like a spherical diamond trying to cope with the climax of a symphony near the label of an LP!

I shall have to select identical passages of each sample for a side by side comparison, as we soon get acclimatised to a given environment. But I shall need time and quiet!

June 25, 2012 at 07:44 AM · I still need to ask the question - if CD quality at 16 bits and 44.1 kHz is the best - why bother with anything less? (Preferably even have 24 bits at 44.1 kHz or an even higher sampling rate - which is how most recordings are made these days ...)

The one sample I downloaded (F) sounded rather poor and muddy to my ear, and it is correct to say that the low mp3 rates do not only limit the higher frequencies but lose detail in other areas too.

June 25, 2012 at 08:09 AM · Mobile devices do not come (yet) with terabytes of capacity. To reduce the size of sound files just to fit more is an issue for some of us.

There are almost no real flat rates for mobile devices, you get full speed for 500MB-1.5GB traffic per month. After that your provider will throttle your download speed. Remote listening gets more affordable with smaller files.

You want to be a "good citizen" and share your sound files with (some of) your friends. So you want to keep their communication cost low.

Wav files need quite some bandwith for remote playing as you can see with my test page. There is almost no way to implement non-stuttering streaming this way.

Most vendors sell downloadable music in mp3 format or alike. So you might want to know what reduction in quality they sell.

So wouldn't it be nice to know, what quality level of mp3 files is acceptable to you to address some of these issues?

And there could be a slight chance (therefore my test) one cannot sense the difference between original and "good" mp3 compression. Why bother with bulky wav files then?

FMF

June 25, 2012 at 09:14 AM · I'm not denying that these are not all valid points. However, on a personal level I never listen to music on the move because I want a quiet environment to listen to classical music. It's probably OK if it's rock or pop which is always (very) loud all of the time, but in the high street, on buses, trains and airoplanes the ambient noise levels are so high that it makes listening a waste of time. (We have cattle trucks for trains and busses here in the UK, and I've never yet been on a plane that has levels below about 80dB.

If I pay for a download of a recording I either purchase the studio quality version, or to save a bit of money and download time, the FLAC version with is close to the wave file in quality and slightly smaller as well.

I think personally that the lower mp3 versions are about the same or worse than the old cassette tapes we used to suffer, and although I am not sure about the highest mp3 quality, I'm sure that the .wav files can't be that much larger?

June 25, 2012 at 09:28 AM · Point taken. I shall publish the sizes (also compared to flac) when the test is finished.

Unfortunately we have 110 visitors so far but six votes only. So the poll result might be less than representative.

FMF

June 25, 2012 at 10:23 AM · Depends on the walkman - the Sony Pro Walkman was pretty good, I knew someone that used it to record a violin -> vinyl album.

FMF, some time back you were pushing ogg vorbis in preference to mp3 - are there any oggs in the sample? And if not why now the mp3 preference? Also, is VBR included or just constant bit rate?

June 25, 2012 at 10:40 AM · "i think a good quality cassette deck is better than almost any mp3 in many ways, # one its analog, clearly superior to digital, its main drawback is noise, which i dont find very objectionable and a slight wow and flutter"

Lyndon - I think this is probably a very personal thing, and although I agree that the very best cassette decks were good (but not as good as a Revox reel to reel at 15 ips!) I still prefer digital.

I do remember returning one reel to reel deck (not Revox) many years ago because of its bad wow and flutter. I think I'm particularly perceptable to any slight variation in pitch, whereas some people can accept it.

I like digital because it's a clean sound and uncoloured (unless you deliberately mess with it) and it has such a low noise floor. So silence is pretty much silence.

June 25, 2012 at 11:44 AM · Nigel, depending on the turnout I am going to provide vorbis and aac tests as well.

However, as already stated, the turnout is pretty low so far.

In this test now you'll find the original as well as CBR, ABR and VBR mp3 files.

FMF

June 25, 2012 at 01:45 PM · There is a significant difference in size between mp3 and CD format wav files: a 320 mp3 file is 22.7% the size of the equivalent wav file, and a 192 mp3 file is 13.6% the size. Note that FMF's files (except for the original) are mp3s which have been saved as wav files, but retaining their mp3 formats, so that it can't be detected solely by inspection of the file sizes which format is which.

Lossless Flac files (which I regularly use for backing up my CDs and old LPs/cassette tapes) are 40 - 60% the size of the original, depending on the quality of the original; a relatively poor quality tape (as regards frequency) will convert to a Flac size of about 40%, and a high quality modern CD will convert to about 60%. This represents a substantial saving of hard drive space.

There are now some decent quality mp3 portables (such as the Archos brand) which also play Flac. Some of the more recent upmarket audio editors also handle Flac and Ogg. The main drawback with Flac for playback on portables, as I see it, is that Flac files are more difficult to tag than mp3. Roll on the day when iTunes provides for Flac!

June 25, 2012 at 02:43 PM · It's actually a difficult question whether to include the original wav or not. As you know there is no such thing like DDD recording. The first link in the chain is the mic which has to "cycle" with the sound - in a purely analog way. So each mic has its own, never flat frequency reponse. Now as we mix several mic signals with different frequency reponse curves the sound engineer ALWAYS brings his (female sound engineers are virtually non-existent, unfortunately) own taste etc. into the equation.

Tests I have done with soloists, orchestral leaders and alike now show that sometimes these "great ear people" prefer even less than perfect mp3s over the original because they feel e.g. the original is "too treble heavy" etc.

And we should never forget that the quality difference between the actual performance and the "original" is so big most of the time that we should really not worry too much about compression quality.

This also explains the (theoretically) non-sense discussion whether LP sound is better than the same recording on CD. If you are used to and liked the LP sound characteristics then you will like everything close to it. And again, the difference in sound between a good turntable and a CD player is so minuscule compared to the huge gap between the live performance and the recording (good acoustics and seating provided). This discussion has significance for people who live in the world of classical recordings rather than of actual performances.

Now move to the world of Electric Light Orchestras (Madonna and alike) and things get less and less significant as there is no such thing like an actual performance anymore. There we have rather "electrical" performances. Can be a form of art, too. A pop star relates to the work of art like the paint manufacturer to Rembrandt. Yes, the artist (the pop star) can tell Rembrandt what he/she likes and how he/she likes it, but it's Rembrandt how actually does it, knows how to do it and puts his own "frequency response" curve into the work. Marketing then decides who will be positioned as the artist, most of the time it's the dancing, moving, glitzy paint manufacturer and not the guy on the mixer controls or the best boy who finishes the work.

FMF

June 25, 2012 at 03:04 PM · FMF

"It's actually a difficult question whether to include the original wav or not. As you know there is no such thing like DDD recording. The first link in the chain is the mic which has to "cycle" with the sound - in a purely analog way. So each mic has it's own, never flat frequency reponse. Now as we mix several mic signal with different frequency reponse curves the sound engineer ALWAYS brings his (female sound engineers are virtually non-existent, unfortunately) own taste etc. into the equation."

I would agree with you. Unfortunately there is a tendency to use Cardiod condenser mics a lot,and sometimes omni condenser mics. (Either small or large diaphragm). This often results in a high frequency emphasis and a loss of smoothness in the highs, and a slightly thin bottom end in some. So condenser mics can and often do sound brittle high up. Other mics are designed to deliberately colour the sound as well in other areas such as mid frequencies.

For this reason I use ribbon mics which are much smoother and less coloured and have a good bottom end. But you need aclean mic preamp for this to work well. These mics are excellent for all strings and piano. (Also brass and woodwind too!)

"Tests I have done with soloists, orchestral leaders and alike now show that sometimes these "great ear people" prefer even less than perfect mp3s over the original because they feel e.g. the original is "too treble heavy" etc."

It is true that a lot of musicians in the classical field do complain about the recorded sound, due to problems with balance, equalisation, emphasised top end, and loss of clarity too.

"And we should never forget that the quality difference between the actual performance and the "original" is so big most of the time that we should really not worry to much about compression quality.

This also explains the (theoretically) non-sense discussion whether LP sound is better than the same recording on CD. If you are used to and liked the LP sound characteristics then you will like everything close to it. And again, the difference in sound between a good turntable and a CD player is so minuscule compared to the huge gap between the live performance and the recording (good acoustics and seating provided). This discussion has significance for people who live in the world of classical recordings rather than of actual performances."

Yes, musicians and studio managers, engineers, producers and the like, must always hear first of all what the real live sound is like, and then try and reproduce that as accurately as possible. Many recordings from the distant past did this very successfully as well as good digital recordings of recent years to the present day.

"Now move to the world of Electric Light Orchestras (Madonna and alike) things get less and less significant as there is no such thing like an actual performance anymore. There we have rather "electrical" performances. Can be a form of art, too. A pop star relates the work of art like the paint manufacturer to Rembrandt. Yes, the artist (the pop star) can tell Rembrandt what he/she likes and how he/she likes it, but it's Rembrandt how actually does it, know how to do it and puts his own "frequency response" curve into the work. Marketing then decides who will be positioned as the artist, most of the time it's the dancing, moving, glitzy paint manufacturer and not the guy on the mixer controls or the best boy."

Yes, that is absolutely right.

Can't go on agreeing with you like this!! (wink)

June 25, 2012 at 04:38 PM · So we have got 8 (EIGHT!) votes so far. Keep up the work, please!

FMF

June 26, 2012 at 05:25 AM · Here is the quickest way to get to your "best" recording for the vote:

compare A to B for as many seconds or minutes you need to pick the "better" one

now compare the "better" one with C for as many seconds or minutes you need to pick the "better" one

now compare the "better" one with D for as many seconds or minutes you need to pick the "better" one

etc.

Because the goal of this test is not to rank all files, but to find your single best. And if you cannot decide with certain pairs then to find the "single" bestS. In this case pick the last of the bests you've found. And drop me an email with all your bests if you like. I shall certainly respond to it.

FMF

June 26, 2012 at 03:59 PM · That's like the standard procedure for selecting the most suitable bow from a bunch of 6, when price isn't the consideration.

June 26, 2012 at 06:53 PM · Do we have a winner yet?

June 26, 2012 at 08:03 PM · You may click on "view results" at any time to monitor progress of the poll.

FMF

June 28, 2012 at 05:59 AM · Did you listeners discover the "human factor" around 2:34? ;-)

FMF

June 28, 2012 at 12:52 PM · I would expect the results to be a bit random, depending on who's taking the tests, and what they're listening with. For instance, there are already tests out there showing that some people PREFER the quality of some types of MP3 because of the way it can change the tonal content of the sound, adding a bit of texture and brightness that mimics presence. Also, I have quite a range of listening equipment, and on the stuff I have of the quality that most people would use, most audio sounds like garbage, regardless of the source file. It's only when I move up the line to better systems that I bet few folks here use that differences start separating.

I rip everything to 320 and can't hear the difference from most original CDs (which these days are generally bad, anyway) on most days. At the most, the difference is meaningless to my enjoyment. At 128, it all sounds like it's being played through a waterfall to me. Somewhere in between is the tipping point for ME and MY situation, only.

Some people only listen to equipment, some to music. I am one of the latter.

June 28, 2012 at 01:50 PM · "Some people only listen to equipment, some to music. I am one of the latter."

Yes this is how I feel too as I am someone whose ears ring constantly (tinitus started around age 35, and seems only to get just a little worse every time I catch a cold).

There are so many older recordings, and how can we enjoy these if the only thing that matters are "statistics" like noise, frequency range, dynamic range, and total harmonic distortion. To the "audiophiles" probably much of Heifetz and Horowitz is intolerable.

June 28, 2012 at 04:12 PM · Michael, I suspect the tipping point for acceptable mp3 listening lies at 192kbps, or possibly 224. Below that, I don't like it, and to my ear 128/160 is only acceptable for the spoken word on the radio, which has its own raft of distortions. Between 192 and 320 I'm not at all sure I can tell the difference, and this is supported by my little research project I detailed a few posts ago.

Your comment that the better levels of mp3 can add something useful to the sound is an interesting one. In accordance with others' explanations above, even the "best" recording can never sound like the original, for unassailable technical reasons. Even if you were there at the time, after the event you have only your imperfect recollection of that live performance on which to base any future comparison.

June 28, 2012 at 04:26 PM · I record concerts here in our shop, and I'm able to monitor either the concert or the recording. There's a real difference, and I see that mirrored in commercial recordings, too. I've derived a nice setup that gives good recorded results in our specific situation, but it can't ever be what the ears hear.

I have one customer with a famous instrument who has been trying to get "the sound" the instrument had on recordings in the 40s. There's no way he's going to get it, because that was never what the instrument sounded like live. I believe that taste in instruments is being skewed by people's constant listening to CDs, and this is not a good thing, since the instrument they like based on CD sound will be shifted even farther in that direction when/if they get to record it, and I don't think they will like where this takes them.

Even the best recordings have certain biases that you can't get around and still sell CDs--the unnatural prominence of the soloist, for instance, and a certain amount of added brilliance that we have come to expect in a recording.

July 2, 2012 at 04:19 PM · *poke*

It really doesn't matter that much, as I have been compressing my audio files (AAC 256kbps VBR) for years and I'm not likely to take all 43 DAYS worth of music in my iTunes library and re-encode it as FLAC.

But,

I did spend an hour or two listening, comparing, repeating, eliminating all of your samples until I found the one that sounded best to my ear. I can see which sample got the most votes (not the one I chose), but where are the details? Which was which, and was I justified in dismissing immediately a couple of the samples?

July 2, 2012 at 10:04 PM · Here are the parameters, bit rates, votes of the tested audio files: http://bit.ly/MpVLcN

It's interesting that compression levels above 160 kbit/s got almost 3x as many votes as the original.

Only "I" and "L" were encoded with constant bit rate. All others except "D" with variable bit rate; "D" was encoded with the more controlled average bit rate.

Another remark: compression level "D" did not get a single vote. And it's widely used in Amazon's downloadable mp3 files.

Thanks a lot for participating!

FMF

July 4, 2012 at 01:46 PM · I recorded some music in CD quality wav format the other day from BBC Radio 3 (on cable broadband) and out of curiosity checked the frequency response in the spectral view. It clearly topped out at between 15500 and 16000 Hz for the whole broadcast, with nothing above. That leads me to think it would be more efficient use of file size when making future recordings from that source to use a sampling rate of 32KHz instead of the standard CD quality 44KHz (sampling rate is twice the frequency response).

July 5, 2012 at 04:34 AM · Interesting!

In the end I decided not to boost the high frequencies to compensate for my ageing ears, but I used my Sony MDR7506 phones which accentuate (4 and 10 kHz) fizz, rosin-noise etc.

I compared very short sections, to avoid getting too accustomed to each sample:

- the opening "silence": the low rumble had a curious warbling at the two lowest bitrates;

- the opening solo: for purity of tone;

- the opening duo: for possible inter-modulation;

- a loud tutti later on: for congestion and saturation;

- in general, I listened also for the quality of the reverberation.

I chose "B". "H" (wave) was my No.3, and "D" my No.2! How come?

Perhaps I actually prefered a fraction less "brilliance", like blurring the pixels of a photo? Perhaps a fraction less clarity seemed less "clinical"? And 44kHz is "only" double the limits of (very young) human hearing..

Many people prefer the sound of an LP to its commercial transfer to CD; I even heard of folks prefering a cassette copy of a CD - the background hiss made it more "human", like the background noise in a concert from breathing and from the movement of clothing.

I watched some my old 8mm film footage: much less detail than digital (amateur) filming, but the "impressionistic" random distribution of colours seems kinder on my eyes.

Thanks for this fascinating and instructive test!

I shall be a little less snobbish in the future, but avoid the common 128 kB/s..

July 5, 2012 at 06:18 AM · Adrian, I've observed a corresponding phenomenon with audio interfaces. There's a manufacturer Metric Halo that produces the ULN-8, not cheap at around €5k for an 8 channel preamp/converter. Stands for ultra low-noise. Personally I love that clarity (several of my works have been recorded with one) but I know musicians that find the sound too pure and clinical, and actually prefer something more down-market.

July 5, 2012 at 06:30 AM · Now, here comes my very personal recommendation (as far as mp3 is concerned):

Use lame -V3 (or another encoder at similar variable bit rate) and you will be on the safe side unless you use extreme upper-end equipment for listening. New headphones under 200$ are not extreme upper-end equipment in this context, even most of the over 200$ headphones will not fall into this category.

FMF

July 7, 2012 at 12:31 PM · This is funny. The results of the vote are statistically significant: they would be unlikely if each of the files had an equal probablity of being voted for. Because the table is sparse, the test is not totally straightforward. Doing a likelihood ratio test with a simulated distribution as a reference, I got p=0,44%.

Now for the funny part: my own preferences were E, H, and I, and I took that as evidence that for my ears, only lame -V0 would be good enough. No test necessary for this n=1 result. When the message is flattering, I'm gullible!

This discussion has been archived and is no longer accepting responses.

Facebook Twitter YouTube Instagram Email

Violinist.com is made possible by...

Shar Music
Shar Music

Yamaha Silent Violin
Yamaha Silent Violin

Pirastro Strings
Pirastro Strings

Find a Summer Music Program
Find a Summer Music Program

Dimitri Musafia, Master Maker of Violin and Viola Cases
Dimitri Musafia, Master Maker of Violin and Viola Cases

Violinist.com Business Directory
Violinist.com Business Directory

Violinist.com Guide to Online Learning
Violinist.com Guide to Online Learning

Dominant Pro Strings

Antonio Strad Violin

Bay Fine Strings Violin Shop

Bobelock Cases

Fiddlerman.com

Fiddlershop

Los Angeles Violin Shop

Nazareth Gevorkian Violins

Violin-Strings.com

Metzler Violin Shop

Leatherwood Bespoke Rosin

Warchal

Barenreiter

Johnson String Instrument and Carriage House Violins

Potter Violins

String Masters

Bein & Company

Annapolis Bows & Violins

Laurie's Books

Discover the best of Violinist.com in these collections of editor Laurie Niles' exclusive interviews.

Violinist.com Interviews Volume 1
Violinist.com Interviews Volume 1, with introduction by Hilary Hahn

Violinist.com Interviews Volume 2
Violinist.com Interviews Volume 2, with introduction by Rachel Barton Pine

Subscribe