 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
98.210.169.208
This is a depressing bit of news.
Follow Ups:
Are iPods changing our perception of music?It's not just iPods..... I think digitized audio in all forms has changed our perception of music.
Are the sounds of MP3s the music we like to hear most?
MP3 is popular because it's most-forgiving to mediocre production and/or mediocre playback. The problem with CD is it only sounds good on a small percentage of audio systems. (And if you go even further, I've yet to hear high-rez sound good on any system.)
Jonathan Berger, professor of music at Stanford, was on a panel with me at a meeting of the American Academy of Arts and Sciences in Mountain View, CA on Saturday. Berger's presentation had a slide titled: "Live, Memorex or MP3." He mentioned that Thomas Edison promoted his phonograph by demonstrating that a person could not tell whether behind a curtain was an opera singer or one of Edison's cylinders playing a recording of the singer. More recently, the famous Memorex ad challenged us to determine whether it was a live performance of Ella Fitzgerald or a recorded one.
If it's just one vocal, even the Edison cylinder can sound decent. The difference shows up when playing a large ensemble or symphony orchestra, where not only capturing all the musicians becomes difficult, but the dynamic scale involved.
Berger then said that he tests his incoming students each year in a similar way. He has them listen to a variety of recordings which use different formats from MP3 to ones of much higher quality. He described the results with some disappointment and frustration, as a music lover might, that each year the preference for music in MP3 format rises.
This is because the quality of recordings has gotten worse, and the artifacts associated these poorer recordings (caused mainly by new storage formats and conversions, creating new sonic problems) are masked with lower resolution playback.
In other words, students prefer the quality of that kind of sound over the sound of music of much higher quality.
The term "quality" should be replaced with "resolution"..... Revealing flaws in the recording isn't necessarily "quality"..... (A quality amp reproducing a unlistenable source will likely sound worse than a lesser-quality amp reproducing that same source. Bose speakers are popular because they make awful upstream electronics tolerable.)
He said that they seemed to prefer "sizzle sounds" that MP3s bring to music. It is a sound they are familiar with.
MP3 doesn't make sounds "sizzle", but masks the "sizzle"..... The "sizzle" is the sonic attribute for today's poorer-sounding recordings. And the masking of the "sizzle" is what the listeners like.
I remember wondering what audiophiles were up to, buying extremely expensive home audio systems to play old vinyl records. They put turntables in sand-filled enclosures with elaborate cabling schemes. I wondered what they heard in that music that I didn't. Someone explained to me that audiophiles liked the sound artifacts of vinyl records -- the crackles of that format.
That's the stereotypical explanation, but not the real one.....
Audiophiles use the "sand-filled enclosures" and "elaborate cabling schemes" to be able to hear more of what was on the vinyl. If audiophiles actually liked the noise artifacts of vinyl, why do they constantly try to minimize them? (Record cleaning machines, antistatic devices, etc.)
It was familiar and comfortable to them, and maybe those affects became a fetish. Is it now becoming the same with iPod lovers?
Since the premise was false, the question becomes useless.......
Our perception changes and we become attuned to what we like -- some like the sizzle and others like the crackle.
Once again, a false conclusion based on false premise......
I wonder if this isn't also something akin to thinking that hot dogs taste better at the ball park. The hot dog is identical to what you'd buy at a grocery store and there aren't many restaurants that serve hot dogs. A hot dog is not that special, except in the right setting.
I think customer feedback on the hot dogs play a bigger role than setting. (AKA placebo effect.) The feedback improves the product. Hence the best-tasting hot dogs possible are often served at the ballpark.
The context changes our perception, particularly when it's so obviously and immediately shared by others. Listening to music on your iPod is not about the sound quality of the music, and it's more than the convenience of listening to music on the move.
Now I agree with that, but earlier, the author cited the "sizzle" for the satisfaction.
It's that so many people are doing it, and you are in the middle of all this, and all of that colors your perception.
That's a factor too..... But with me, since I do prefer WAV files over MP3, that's all I listen to on an iPod......
All that sizzle is a cultural artifact and a tie that binds us.
Maybe it's peer pressure. The single most-destructive internal force in society, in my humble opinion.
It's mostly invisible to us but it is something future generations looking back might find curious because these preferences won't be obvious to them.
Because "social relevance" has become the yardstick for present music instead of the quality of the music product at face value. This is why the Beatles will be appreciated 50 years from now, but not Linkin Park.
On a related note, a friend commented recently that she doesn't understand why people put up with such poor sound quality for phone calls on cell phones, and particularly iPhones. "I can hardly hear the person talking to me," she said. "I don't think smart phones are making any improvement to the quality of the phone call," she added. "Is it not important anymore?" She wondered why people accepted such poor quality, and so did Jonathan Berger, but a lot of people just don't hear it the same way.
It might be a cold day in Hell before a cell phone sounds like a good land line. The technology, in its current state, is still too limited to attain such quality. Unless a breakthrough technology transforms audio quality via wireless transmission (in which RFI is a major hurdle), this issue will be sticking with us indefinitely.
Edits: 03/11/09 03/11/09
I've said this before. LAME and presumably other MP3 encoders roll off frequencies higher than 19 khz. This is probably an improvement for the vast majority of users of ipod-generation types of equipment.
I'm 23 and when I turn up the 19K band in an equalizer I can hear it. Maybe it's just lower harmonics but I downloaded a sine wave at 19kHz about two years ago and could hear it (it sounds absolutely awful). I couldn't hear 20kHz, either because my hearing isn't that good or because my sound card couldn't reproduce it (they're typically 20-20,000Hz, right?).
I greatly prefer my stereo to other stuff. I'm not a snob in that I'll listen to inferior systems and enjoy the music, if not quite as much, still a great deal. I love my Discman on trips (although I guess it's just called Walkman now, but I still associate that with tapes). An MP3 player seems awfully expensive for something I'll use so rarely so I haven't got one (plus I'd have to re-encode all my digital music files to MP3), but I use my Etymotic 4P in-ear headphones with my portable and it sounds fine. I just prefer my home stereo and I've loved high-quality sound reproduction since I got into it when I was 20 or so. I haven't got a great or expensive system, and I don't see being in the financial position to upgrade for several years yet, but tubes + a good pair of speakers + CDs or FLAC files = bliss.
I am kind of old-fashioned in that I still buy albums on CD instead of downloading MP3s. I don't want to spend money and get nothing physical. I think I'm the very last year or so of the pre-iPod generation.
There are some benefits to being young. Hearing high frequencies is one of them.
When I was three years younger than you, I was able to hear 21,000, where the signal came out of an audio oscillator. Those days are over 40 years gone. Now I don't hear the 15,750 horizontal flyback sound from the one remaining CRT TV in my house. :-)
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
That was the pivotal rude awakening for me, suddenly one season I couldn't chase them down by ear. But I think good tweeters are still important, my hunch is that the transient detection mechanism lasts longer than the continuous detection scheme.
Rick
Actually, a lot of MP3 encoders roll off at 16kHz. The latest version of LAME rolls off at around 19kHz only if you use the highest quality setting (-v0).
You are right in that the roll off could well cause a subjective improvement in quality. For one thing, it would get rid of all the noise above 20kHz dumped by noise shapers like UV22 (used in a lot of recordings).
Not that we can hear it anyway - hands up those who failed the 20kHz listening test posted a few threads ago! :-)
Somebody might hear who even failed the test. In fact, most normal people younger than I can hear it. Hearing it by itself, and distinguishing whether it is there in the context of other sounds, are different issues.
with most college students speaking in slang and not knowing how to read, it's no doubt they can't hear either. maybe there is something in the soda pop?
.
Thank God.
I remember a test a few years which had quite a few "golden ears" (including some recording engineers) judging MP3 files to be "superior" to their uncompressed counterparts.
Our ears seem to be easily fooled into thinking "different" = "better."
MP3 artifacts can be similar to the noise suppression artifacts used in some audio cleanup software. If there is some "information" that is deemed by the MP3 encoder guru's to be "unimportant" or "inaudible" it will be thrown away to save bandwidth. Cleanup software also throws away portions of the frequency spectrum in blocks of time, when it considers that these blocks contain unmasked noise and not music.
It is possible on some recordings that the information being thrown away is actually noise or distortion and not part of the music. Throwing it away can therefore actually improve the sound of some recordings. Audio cleanup software gives the user some degree of control over what is happening, so there is a good possibility of sonic improvement on some damaged or older recordings, but only when this software is used judiciously by a skilled operator. MP3 compression doesn't have this degree of user control and so is much less likely to improve the sound. But I have heard cases where it successfully removed noise and distortion from some bad recordings.
Given the square waves created by many modern pop recordings, I would have no trouble believing that these would be improved sonically by being compressed. Only I would go a bit further. I could compress them all the way down to 0 bits per second using the power switch.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
*** Given the square waves created by many modern pop recordings, I would have no trouble believing that these would be improved sonically by being compressed. ***
Actually, based on my casual observations, the opposite is true.
Take an uncompressed recording, say of a piano solo. Convert it to MP3 using variable bit rate (I recommend LAME at say -v 2). Or maybe Ogg Vorbis at a specific quality level (say 0.6).
Now apply a peak limiter. Convert the peak limited version using the same parameters.
You will find the latter results in a higher bit rate than the former, indicating more bits are required to maintain the same level of MP3 quality.
Interesting, isn't it?
I suspect dynamic compression makes the signal harder to compress - the shape of the transients become more rounded and therefore require more bits to replicate.
I think we are talking about different definitions of "quality". By quality I was not talking about the setting used with an MP3 compressor (e.g. quality parameter for VBR). I was talking about perceived audio quality.
As to the question of why more bits are used with clipped or distorted recordings, I have some experience rendering many dozens of recordings into MP3, FLAC, Apple Lossless and WMA lossless. Recordings definitely vary tremendously in the amount of compression that is possible. On occasion I have looked into this and investigated how these algorithms work in some detail.
Starting with your MP3 piano example, what is probably happening is that the peak clipping is adding high frequency content that is missing from most portions of the recording. The distorted version simply has more information in it than the filtered version, and hence does not compress as well.
This also happens with lossless compression. Lossless compression works by predicting the next sample from previous samples and then transmitting only the difference. If prediction is good, then the differences will be small, and these can be encoded with few bits. If prediction is poor, the differences will require more bits. Most lossless compression uses linear prediction, which won't be able to predict non-linear signal artifacts such as those created by clipping or complicated non-linear devices such as compressors and limiters.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
*** I think we are talking about different definitions of "quality". ***
I think you misunderstood the point I was making.
The statement you made originally was
*** Given the square waves created by many modern pop recordings, I would have no trouble believing that these would be improved sonically by being compressed ***
What I was trying to show was that this belief is not correct. Dynamically compressed recordings will suffer more perceived "damage" from being lossy compressed (at a particular bitrate) compared with the equivalent non-dynamically compressed recording.
I know you were being somewhat flippant and you were probably trying to say "it's so damaged anyway it can only get better." However, I don't agree. I know these days it's fashionable to deplore the amount of compression used in commercial recordings, and yes we've all seen some truly horrendous examples.
But I don't believe non-linear dynamic compression is necessarily evil. I tend to apply it to my own recordings, though not nearly as aggressive as say TV broadcast standards. It's kind of impossible to get that "commercial" sound without using some form of compression, especially on vocals and drums. In other words, the technique is not just used to "make the music louder" - it's an intentional sound effect layered onto the music.
Of course you can say "I hate that kind of music/sound" and certainly it is your prerogative. But some people do like it. Certain types of music require that kind of sound - it's as much part of the music as the notes and the lyrics.
As you correctly pointed out, dynamic compression introduces additional high frequency harmonics into the signal. These harmonics are essential to the overall "sound" of the recording. These additional harmonics actually make it harder to compress, and therefore an MP3 version of a dynamically compressed recording is less likely to sound "better" than the original (if you take the view that the compressed "sound" is the intent of the engineer, and these additional harmonic content is essential in preserving the nature of that sound).
Yes, it is part of commercial sound to compress music. And it does make it louder. In small amounts it can creates a "house sound" and is suitable for certain types of music. Compression makes it possible to listen at louder volumes on cheap equipment. Carried to a limit you get a sound which radio hams used to call "communications quality", a good way to get voice through noise when a radio transmitter is low powered. (Applied to symphony broadcasts on FM the result can be a disaster.)
I don't believe there are any general rules that can say what happens when you apply MP3 encoding to music that has had its loudness increased as part of a mastering process. I suspect the possible interactions are many and complex. I have seen cases where MP3 encoding improved sound quality, and others (the general case) where it made matters worse. Same with required rates for VBR encoding at a given quality level. The interaction of complex non-linear systems are not easily analyzed.
In the thousands of classical recordings I own, I have come across very few where I felt there was too much dynamic range and my listening experience would have benefited from compression. All it takes is a twist of the volume control and an amplifier and speakers properly sized to the listening room to play back music at the proper levels.
I don't listen to pop music, so I am not qualified to comment on how much compression is appropriate to this class of music. More I could say, but it would be impolitic to comment on the relative merits of various musical genres in this forum.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
I used to be against compressing classical recordings, but now I think in limited amounts it would probably be beneficial to all but the best systems.
You are right in that the trick is to move the average volume into the "sweet zone" - loud enough that it is clearly distinguishable from the background noise, but not so loud that the peaks are completely squashed.
Another benefit of peak limiting is to prevent the possibility of loud passages hard clipping.
One problem with non compressed recordings (particularly of symphony orchestras) is that the peak to average ratio is extremely high: +30 to +40dB is not unusual. If we say the average level is 80dB SPL, it means peaks are as high as 110-120dB SPL, which will surely cause an amplifier to clip unless you have thousands of watts of amplification (or extremely efficient speakers).
For example, my setup, which is bi-amped (250w RMS x 2 on each front speakers, total of 750w RMS for all three speakers and 6 drivers). Even with this massive amount of amplification (the amplifiers require their own dedicated 20A power supply) I have calculated that clipping will occur past 115dB SPL. I imagine a more typical setup would clip at far lower levels.
Of course, one could always listen at a lower average level, say at 60-70dB SPL, which a lot of people do. But the problem then is that the average level is too close to the ambient noise level (typically around 40-50 dB SPL), which is why the "reference" listening level for monitoring is recommended as 80dB SPL.
I've noticed that on most classical recordings some form of peak limiting have been applied, typically to bring down the peak to average ratio from over +25 dB to something like 15-20 dB.
I think this is perfectly acceptable and will make the music less likely to clip on most systems.
Of course, the non compressed version is better if your system can handle high SPL with no clipping.
Maybe the best solution is to implement peak limiting during playback, and leave the recording uncompressed. You could potentially flag all recordings with appropriate peak limiting parameters for "safe reproduction." That way, a player will decode the parameters and automatically apply peak limiting to prevent amplifier clipping, and the amount of peak limiting can be tuned to the capability of the playback system.
Dolby Digital dialogue normalization is one such implementation. Unfortunately, the normalization is optimized for speech, not music.
I checked two hi-res recordings of Mahler 6th, one that I just downloaded from the Boston Symphony and the other the Acousence recording. For the first five minutes the peak to average ratio on both recordings was 23 db. I think this is typical of most classical recordings and represents the normal dynamics of orchestral performance. Looking at the loudest one second portion, the peak to average ratio was around 12 db. In my case, I use 90 db efficient speakers driven by 25 watts each. I listen in near field, 1 m from the speakers so I can get average SPL of 90 db from these recordings. With both recordings, I (just barely) have enough head room and that assumes that I turn off the computer that I am presently using to compose this post so the quiet portions are not disturbed by room noise. So far as I know these recordings aren't compressed. Both are extremely dynamic. They would both be completely unlistenable in an automobile.
On my web site you can find my wife's piano recitals. If you look at the dynamics of these you will find that a typical peak to average ratio for a 20 minute piece is about 20 db, and looking at the loudest 2 seconds around 12 db. There is no compression on these recordings other than that from the analog master tape itself (recorded 2 track at 7.5 ips).
I have a version of the Boston Symphony Mahler 6 that was taken off of FM from WCRB, which has a notorious compressor. For the entire test period the peak to average ratio is 13.5 db. For the loudest 1 second the peak to average ratio was 10 db. This processing sapped the life out of the music but only saved a little in amplifier power if the goal is realistic loudness. It would be preferable to listen to this version in an automobile, however.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Here's two of my non-piano recordings that I was able to find at short notice:Sydney Symphony Orchestra, recorded at the Sydney Opera House concert hall (practice session, preparing for a concert) - around 32dB peak to RMS
Adrian Cunningham Quartet, recorded at the Credo restaurant during the Thredbo Jazz Festival 2007 - around 26dB peak to RMS
The former was recorded from the upper stalls, so quite far from the orchestra. Typically the SSO would register +35-40dB if recorded from row F.
The latter was recorded quite close to the band, using a portable field recorder, so the mic wasn't very good. Matt Baker, the pianist, promised to get me a feed from the soundboard prior to the performance, but due to technical difficulties that wasn't possible (not all instruments were miced into the board).
On the plus side, we were seated pretty much at the optimal position. I met the restaurant owner the night before, and he gave me the best table when he found out I was recording.
Edits: 03/10/09
Perhaps the microphones on your field recorder have better frequency response in the ultrasonic range than were used on my recordings.
In addition to the recording of Beethoven Op 101 made in Jordan Hall (microphones in the usual position on an Ampex 350 series machine by the New England Conservatory audio department in 1975 and 1977) I have a recording I made in my home on my wife's Steinway B. A pair of AKG C-451E cardiods was placed just inside the open lid of the piano and recorded at 7.5 IPS on a 2 track Tandberg 6021. This recording was digitized at 96/24. For the entire sonata the peak to RMS ratio is 23 db. For the loudest 2 seconds that contains a highest peak, it is 14 db. This recording was made with very close miking to deliberately eliminate room reverberations. When played back on Snell A/IVs there was no significant difference between the piano and the reproduction, and with a 150 watt/channel amplifier there was a few db of headroom left prior to lighting the soft clipping lights on the amp. (It took much experimentation to achieve this result, and the recording doesn't sound particularly good in other rooms.) Perhaps I am getting a slight amount of compression from the tape, but the levels were kept down to eliminate distortion.
I also looked at another recording that I made at an excessively live room during a yoga retreat at which there were singers, acoustic guitars, harmonium, and various percussion instruments. For logistical reasons, the microphones were hung high from the ceiling and the resulting recording had excessive reverberation. The same AKG microphones were used and were recorded digitally at 44/16 using a Zoom H4. This resulted in a 23 db peak to average for 50 minutes of recording. It appears from looking at the recorded peaks that a small amount (less than 3 db) of compression was added by the H4 prior to digitization with this recording. By looking at some songs which were quiet and had no peaks above -6 dbfs the peak to average ratio was 23 db.
A similar result was made in a different recording using the H4 Zoom's own microphones. (This recording is for sale on Innersong.) My measurements are taken from the contents of the H4 memory stick. When mastering the album it was necessary to add a small amount of compression to the songs that had been recorded without invoking the compressor to make the recordings made on several successive days have a similar sound. Most of the songs were released at a peak to average ratio of 17 db. Personally, I would have preferred if there had been no compression in the original recording, but the recordings have to sound OK on boom boxes, etc.—these songs are marketed to yogis, not audiophiles.
I do have one recording where the peak to average ratio would have been more than 30 db, but the original was completely clipped. It was made in an open tent during a windstorm and at one point the tent collapsed, knocking down the mic stand. Fortunately, these were cheap dynamic microphones and were undamaged. :-)
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Just in case you thought that all my examples were from my portable recorder - only 1 example (the one recorded at Credo) was. The others were done using mics available at the locations - the piano ones were done using Neumannn (can't remember model number).
*** here was a few db of headroom left prior to lighting the soft clipping lights on the amp ***
This is not a reliable indication of whether the amplifier is clipping or not. As you can imagine, short duration clips will not be visible at all.
If you do the maths, a 150w amp on 90dB@1W1m efficiency speakers *will* clip playing back uncompressed recordings with the peaks that I quoted (assuming you are listening from at least 2m away). If you don't believe me stick a CRO on the speaker cable and watch those flattened peaks.
Even my system will clip for peaks greatly exceeding 115dB SPL, and I have a total of over 2000 watts of RMS amplification deployed across 7.1 speakers. A good studio mic will capture up to 130dB SPL, so even I can't play back uncompressed recordings without clipping.
Perhaps you can send me some sample recordings. I would like to see these peaks! If you are interested, email me and we can set up FTP access.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Piano: Steinway Model D, recorded in a medium sized hall
Recorded at 24 bit 44.1kHz (for eventual burning onto CD)Values are Peak to RMS, expressed in dB:
Beethoven Sonata No. 23 (+23dB)
Chopin Etude Op. 10 No. 4 (+21dB)
Liszt Mephisto Waltz (+22dB)
Bach Prelude and Fugue in Bb minor (+21dB)
Beethoven Appassionata Sonata (+24dB)
Gershwin Rhapsody in Blue (+25dB)As you can see, across a varied repetoire, not a single recording was under 20dB.
+30dB is easily encountered in a smaller room with hard reverb and a closely placed mic.
I can assure you a modern symphonic orchestra is capable of much higher dynamics than a single acoustic piano.
Edits: 03/10/09
*** For the first five minutes the peak to average ratio on both recordings was 23 db. I think this is typical of most classical recordings and represents the normal dynamics of orchestral performance. ***
Well, I have recorded orchestras before, and I can assure you +30 to +40 dB is common (at least, for the orchestras I have recorded). In fact, it can exceed +50 dB.
I have also recorded acoustic pianos, and I have definitely captured +30dB for piano performances (including mine), on several occasions. Depends on the hall acoustics though. As for recordings of your wife's playing, remember that analog tape acts as a peak limiter.
Also as a reference, the La Boheme LP I converted to digital last week has a peak to average ratio of +30 dB. In fact, a lot LPs, even non classical ones, have peak to average ratios anywhere from +20 to +30 dB.
"Also as a reference, the La Boheme LP I converted to digital last week has a peak to average ratio of +30 dB. In fact, a lot LPs, even non classical ones, have peak to average ratios anywhere from +20 to +30 dB."
And these are probably the best sounding recordings on a good system. What I fail to understand is why we have uncompressed recordings issued on a medium that supposedly has "limited" dynamic range and then we go to a medium that supposedly has "perfect" dynamic range and need to apply compression.
I think the solution to the playback problem (peak power and room noise) is to release albums in two formats, one uncompressed and the other compressed by the mastering engineer. (These low quality recordings could be released in MP3 format.) In my opinion, this would be a big improvement over the existing situation, where audiophile labels produce high quality recordings of generally second and third rate performances and mainstream labels produce low quality recordings of first rate performances. (There are exceptions.) This situation is changing, as the mainstream labels fold up and musicians start releasing their own material. This leads to an urgent problem: educating the self-producing musicians.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
*** I think the solution to the playback problem (peak power and room noise) is to release albums in two formats, one uncompressed and the other compressed by the mastering engineer. ***
Already done. Many of Pat Metheny's recordings are mastered differently for CD and for playback on radio. Utada Hikaru's Heart Station was mastered with less compression for the CD release than the digital download.
However, I think a better solution is to do peak limiting on the fly during playback, and embed appropriate compression parameters as flags in the recordings. This is implemented in AC-3 as dialogue normalization, as I've described before.
You have proposed a sensible technical approach. It might work in some market segments, e.g. players for use in luxury automobiles. But it would be a hard sell for the mass market because it adds cost to low end playback systems, where volumes are highest and margins are least. Furthermore, there can be economic advantages to a seller for releasing multiple versions at different quality and price tiers. They can segment the market and collect more revenue from rich consumers than from poor ones. While we may not like it, that's how Capitalism works. (Or has worked in the past, some say its days are numbered.)
Creating multiple versions in a shiny disk, warehouse, truck and retail store world is expensive, particularly for low volume titles. However, in an Internet server download world multiple versions add negligible cost, even to low volume titles. Hence the multiple versions and price tiers on HDtracks, Linn, and the Boston Symphony, to name just a few sites. (These sites' price differences are not explained by bandwidth costs, which are pennies even for hi-res.)
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
"It is possible on some recordings that the information being thrown away is actually noise or distortion and not part of the music. Throwing it away can therefore actually improve the sound of some recordings."
To my ears MP3 encoding at 192kbps or higher sounds better than lossless 44.1kHz or CDs as I feel it throws away part of what makes 44.1kHz sound so bad, the stridency and coldness and adds very welcome artificial analog-like warmth. Problem is it throws away too much of the resolution. I can listen to MP3 much easier than lossless uncompressed 44.1kHz files. But I much prefer 24 Bit 96kHz lossless uncompressed music files as they have 'REAL" warmth and no apparent loss of resolution.
The trick will be to get kids to jump from MP3 to 24/96. I think it will happen.
From my new article at PFO:
Part Six: I was wrong and too forgiving of lower resolution PCM
"When I first got iTunes and a iPod a year and a half ago, I was thrilled with the 192kbps and higher MP3s especially the free ones from Download.com. I never liked the sound of CDs but these MP3s were warmer and more ambient with less strident highs. I commented at the time part of what they threw away were the things that made CDs sound so bad ...to me at least.
I learned over time that MP3's frequency response starts rolling off at 16kHz or lower, thus the highs were smoother because they were lower in level. This could also be what caused the extra warmth. It is possible the extra feel of ambiance is artificially created by the lower signal to noise ratio of MP3.
However, over time I just could not live with the loss of resolution caused by encoding to MP3, so I started listening to uncompressed lossless WAV and AIFF music files, and they indeed have more resolution. However there was some of the stridency and dryness of CD, only not quite as bad. I had the iPod shuffle so I cannot do Apple lossless which in my early tests I found warmer than WAV or AIFF.
It seems on my iPod with 44.1kHz, I can either have resolution or warmth and ambiance, but not both at the same time.
I find 24 Bit 96kHz lossless uncompressed music files have much greater resolution than any type of 44.1kHz music files. They have even more warmth and ambiance than MP3, which would be "real" warmth and ambiance instead of artificially created variety used in MP3s. However I couldn't play my 24 Bit 96kHz music files on my iPod, so I sold my iPod, and deleted all computer music files lower than 24 Bit 96kHz. Now on my trips outside my home, I carry no music. I just experience the real world instead. At first I thought I might wait for an acceptable 24 Bit 96kHz portable player, but now I am not so sure I want to have music with me on walks and other travels.
I now recommend NO digital products or music files with resolution lower than 88.2kHz PCM with the sole exception of the Telarc historical 50kHz Soundstream recordings on SACDs or LPs."
Give me high resolution or remain silent,
Teresa
It sounds like you are bothered by ringing artifacts associated with 44/16 recording and playback. Given that you can tolerate MP3, realize that it is 44/16 material with the portions of the frequency range removed according to a complex algorithm. (Some MP3 recordings are 48/16.) If you use straight forward filtering, you could easily remove these artifacts with a filter that would be in operation all of the time, unlike an MP3 encoder which is constantly deciding what information to keep and what information to throw away. This might give you better results than MP3 encoding, because it would be consistent over time, and hence less disrupting to the ambiance. MP3 encoders are designed to do a good job on preserving the tonal quality of music, but they do much less well with ambiance and sound stage.
It is also possible that the high frequency response of your amplifier and speakers may exaggerate the artifacts if they have any resonances or non-linearities around 22,050 Hz. If the artifacts are above the frequencies that you can hear (likely since you are not a teenager) then they are probably affecting your perception through non-linear distortion, either in your system or in your ears. These factors would be most operative with music that has lots of high frequencies played at loud volumes and/or listened close to the speakers. By varying the program materials, volume control, and distance from the speakers it may be possible to tease out which factor(s) most affect you and your system.
On occasion I come across a bad recording of an excellent performance, a recording that is unlistenable at my normal playback volume. In some cases, I can still enjoy this recording by turning down the volume. This is common with 44/16 material, but occasionally happens with some hi-res digital material, especially some analog transfers that come with built-in distortion. Most of my hi-res material does not have this problem. (Note that I am not talking about simply running out of headroom in my amplifier or speakers. This is usually not a problem in my listening room, except on a very few recordings, typically Mahler symphonies.)
If you have an audio editor it will come with editing and mastering tools that will enable you to experiment with various sorts of filtering and sample rate conversions as well as take the spectrum of particular recordings at arbitrary points. In my opinion, audiophiles who have a working computer audio system should seriously consider obtaining such software and play around with it, if only for its educational benefits. Some programs are free. Others come with free trial downloads.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Give me high resolution or remain silent,
Teresa
it may be that throwing away some bits actually improves the sound.
..
Post a Followup:
| FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
||||||||||||||
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2024, Audio Asylum, all rights reserved ]
