Becoming An Audiophile Or: How I Learned To Stop Accepting Sub-320kbps Bit Rates And Love FLAC

[#SnowyVernalSpringsEternal 10,379]

dBpoweramp Music Converter has a batch converter. I'm sure there are others.

[Holko 10,127]

foobar2000, maybe?

[Jay 39,199]

dbPowerAmp is a GREAT batch converter.

[Kühni 484]

.

[Kühni 484]

.

[The Great Gonzales 6,224]

I assume if you are ripping a 16/44 CD, to 24/96 FLAC, that might cause it, but I don't really know, me not really being an expert about that sort of thing.

[Drew 593]

I have all of my CDs ripped as AIFF files (1411 kbps). They are better-quality than FLAC and basically are exact copies of the data on the CDs. I also recommend AIFF instead of WAV because AIFF supports metadata such as album artwork. The only downside is storage on my phone and iPod. iTunes is stupid and only has the option to downgrade higher-quality audio files to 128 kbps to fit more onto devices.

[Jay 39,199]

10 minutes ago, Drew said:

I have all of my CDs ripped as AIFF files (1411 kbps). They are better-quality than FLAC

 

That is incorrect.  They resulting output is 100% identical.  FLAC files simply take up less space.

[Drew 593]

4 minutes ago, Jay said:

That is incorrect.  They resulting output is 100% identical.  FLAC files simply take up less space.

1

 

I'm assuming that FLAC is comparable to Apple Lossless, which uses automatic VBR from around 400 kbps to 1000 kbps. I know that I heard a difference when I switched from Apple Lossless to AIFF. The difference would depend on what bitrate a track was on Apple Lossless. Upgrading from 400 kbps to 1411 kbps would be more noticeable than upgrading from 900 kbps to 1411 kbps, for example.

[Taikomochi 1,213]

FLAC is constant bitrate

[Jay 39,199]

FLAC is lossless.  It's right in the name.

[Drew 593]

Every FLAC album I have come across has been VBR under 1000 kbps.

[Jay 39,199]

You're doing something wrong.  I don't know what to tell you.

There's no quality loss when encoding to flac.  That is literally the entire point of it existing.

[The Train Station 8,601]

Don't give him FLAC about it.

[Drew 593]

1. This is what my computer says. Is the computer lying? 

2. I still don't understand how there isn't any quality loss if the bitrate is lower than the CD's.

[The Train Station 8,601]

Whoever altered the databanks... is here.

[Jay 39,199]

It's not lying. The kbps rate is lower, because, as has been stated, that is the point of FLAC encoding: to make the files smaller.  However, no audio fidelity is lost.  That, again, is the point of LOSSLESS audio compression.

Do you understand now?

[Drew 593]

3 minutes ago, Jay said:

It's not lying. The kbps rate is lower, because, as has been stated, that is the point of FLAC encoding: to make the files smaller.  However, no audio fidelity is lost.  That, again, is the point of LOSSLESS audio compression.

Do you understand now?

 

I understand. I'm just not sure if I agree that no quality is lost because I was able to tell a difference after upgrading to AIFF. Anyway, my main issue is if the FLAC files have a constant bitrate. You guys say they do but my computer disagrees...

[Taikomochi 1,213]

Honestly, I am not the most in-the-know person on this matter. Jay seems to know a lot more than me, so you should probably just discount what I said regarding the bit rate of FLAC

[Jay 39,199]

There shouldn't be any audible difference between an AIFF encode and a FLAC encode of the same source material.

if you hear a difference, there was either a problem with the encoding, or a problem with the playback.

[Drew 593]

6 minutes ago, Jay said:

There shouldn't be any audible difference between an AIFF encode and a FLAC encode of the same source material.

if you hear a difference, there was either a problem with the encoding, or a problem with the playback.

 

Could I just have sensitive hearing? This isn't the first time that people didn't believe I could hear a difference between lossless and uncompressed.

[Jay 39,199]

The DAC in your computer / device could not be properly decoding one or the other

[Taikomochi 1,213]

That’s pretty plausible actually. I’ve played different lossless codecs in MPC and VLC and they badly distort the audio, depending.

[Marian Schedenig 8,900]

16 hours ago, Drew said:

 

I understand. I'm just not sure if I agree that no quality is lost because I was able to tell a difference after upgrading to AIFF. Anyway, my main issue is if the FLAC files have a constant bitrate. You guys say they do but my computer disagrees...

 

No, you don't understand. We're talking about mathematical facts, not about different views or perceptions. I don't know the technical details about the FLAC codec, but I know it's lossless and just by default I would assume it is in fact VBR. That's not a contradiction.

 

If you create a Zip (or Rar or whatever) archive of files on your computer, that archive will be smaller than the sum of all the files you put in it. Because it's compressed. But if you uncompress it, you will get exactly the same files as before (otherwise a program or a text compressed with Zip would be broken after unzipping). That's called lossless compression. It's still compressed, because the compression algorithm makes certain probability assumptions about its contents.

 

Take text files, for example. They mostly consist of upper case letters, lower case letters, numbers and a few additional characters like spaces, line breaks and commas. Let's say those characters add up to about 26 + 26 + 10 + 10 = 72 characters. In a text file, each of these characters is stored in (at least) on byte. A byte can take any value between 0 and 255, but for a text file, most bytes only contain one of 72 different values. You can therefore compress the file by putting the first character in the first byte, along with information about the next character or characters. Then the second byte might start with information about the third or fourth character, etc. You'll end up with a file that describes exactly (!) the same text, but takes up less space. If you apply the corresponding decompression algorithm, it will output an exact (!) copy of the original text file. That's lossless compression - it is in fact lossless.

 

Now if you try to compress a file that contains completely random data (each byte contains a random value between 0 and 255), and the content is so random that you don't have certain values that appear significantly more often than others (as described above for text files), you won't be able to make it any smaller by compressing it. But most files you compress will be text files, or images, or audio files, or program files, or data files, all of which are decidedly non-random. They have values that appear significantly more often than others, and certain repetitive patterns, that can be used to create a smaller representation of the files without losing any information. For example, "AAAAAAAAAAAAAAAAAAAA" takes up twenty characters. "20xA" takes up four characters, but if I tell you to write down the first one and then tell you to write down "20 times A", you'll end up with the same text.

 

Different types of files have different types of characteristics. The value distributions and repetitive patterns in text files and program files, for examples, are decidedly different. So are those for images and audio data. You can compress all kinds of files with general purpose algorithms like Zip and Rar, but if you know that you're going to compress image files, you can choose to use a different compression algorithm which is optimised for the distributions and patterns for images and will therefore be able to compress your image files more efficiently. You could for example use PNG to losslessly compress your image, or JPEG to make it even smaller, although in this case with a loss of certain details. JPEG and other lossy compression algorithms know about the data they want to compress (e.g. images) and decide what information they can throw away without anyone really noticing. For example, if you have a 100,000 x 100,000 pixels image that you're only going to view as a whole, and every pixel has the exact same colour (255,0,0 - a pure full red), but one single pixel somewhere around the middle has a minimally different colour (254,0,0 - a pure, nearly full red) which your eyes won't be able to notice, even if your display hardware is good enough to actually display it), you can take the original image of 100,000 x 100,000 x 3 bytes (3 bytes means 24 bits per pixel), i.e. 28 gigabytes, and compress it to something that says "100,000 pixels wide, 100,000 pixels high, all of them red", which you could fit in 7 bytes (!). That would be a phenomenal compression ratio of 1 to roughly 4 billion. If you want to compress it losslessly, you would have to at least say "100,000 pixels wide, 100,000 pixels high, all of them red, except pixel 45,194/49,240, which is nearly red", which would e.g. take 14 bytes - still tiny compared to the raw data, but twice the size of our lossy version. But for that increase in size, you would get the ability to restore the exact original image from your 14 bytes file, including that special pixel in the middle. That's the difference between formats like JPEG and PNG.

 

And for audio files, you have the same situation. You have lossy formats like MP3, Ogg Vorbis and AAC, and lossless formats like WAV (typically no compression at all), FLAC and Apple Lossless. The former work by purging audio data they believe you cannot hear (for example a barely audible noise playing at the same time as an explosion, or very low or high frequencies that most humans cannot hear anyway, or more complex psychoacoustics) and discard them. That means MP3 & Co can create a smaller file than a lossless algorithm like FLAC can, but you will not be able to exactly recreate the original raw audio data, and if you take away too much information, you will in fact be able to hear the difference.

 

FLAC & Co don't do that. They don't take away anything. They just use certain properties of audio data to make files smaller by saying the audio equivalent of "20xA" instead of "AAAAAAAAAA". Much more complicated than that in detail, of course, which is why it also takes CPU power to do the compression and decompression. You can therefore choose the compression strength - stronger compression means that the files get smaller, but compression (and possibly decompression) takes longer. If you want your compression to be faster, or if you plan to play the compressed files on a device that has a slow CPU (an old mobile phone) or a lot of other stuff going on (a gaming console), you might opt for slightly larger files that need slightly less CPU time on playback.

 

And as for variable bit rates: That just means that the compression makes some parts of files smaller than others. For example, if you have a text like "AAAAAAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAA", you could compress it to "20xA1xB20xA". That's a variable bitrate compression - the 20 As at the beginning and and have been compressed to "20xA", i.e. 4 characters each, while the single "B" in the middle has actually become larger by turning it into "1xB". You have a lower bitrate in the beginning and end than in the middle, but you still have an overall smaller file, and you're still lossless.

[Jay 39,199]

Well said, Marian!  Nicely done!

[Bayesian 1,428]

This seems to be the right thread to bump to let folks know I've decided to give audiophilia a try. To be sure, I'll be taking it slow. Very slow, really, since I have no money to spend on equipment and likely won't for some time. (Then again, I wouldn't know what to look for even if I had the money, so it's probably a good thing I'm going to focus on learning what I need to know and take it from there.)

 

My inspiration comes from a gradual awakening. It's finally clicked with me that it's one thing to obsessively collect the music of a brilliant talent like JW, but entirely another to listen to that music the way the people involved in its preservation on disc -- from JW himself as conductor and the orchestras/soloists/choruses he leads to the producers and engineers who create the final recording and/or locate, assemble, remix and remaster earlier efforts for posterity -- want it to be enjoyed. I'm ready to start learning how to really listen to all the wonderful music I've been collecting.

[Dixon Hill 4,236]

Ye gods I was full of shit.  Probably still am.

[Manakin Skywalker 5,229]

1 hour ago, Bayesian said:

This seems to be the right thread to bump to let folks know I've decided to give audiophilia a try. To be sure, I'll be taking it slow. Very slow, really, since I have no money to spend on equipment and likely won't for some time. (Then again, I wouldn't know what to look for even if I had the money, so it's probably a good thing I'm going to focus on learning what I need to know and take it from there.)

 

All you really need is any decent pair of modern headphones/speakers. There's not really anything else you need in terms of equipment.

[Jurassic Shark 12,902]

On 3/23/2018 at 11:26 PM, Kühni said:

.

 

You can say that again.

 

On 5/28/2018 at 6:49 PM, Kühni said:

.

 

Oh, you did.

 

1 hour ago, Manakin Skywalker said:

 

All you really need is any decent pair of modern headphones/speakers. There's not really anything else you need in terms of equipment.

 

And a decent DAC, either stand-alone or part of your amp or player.

[rough cut 1,742]

@Bayesian Good for you! Moving from mp3s to lossless really makes a difference. As do a good audio system! 

 

As for files and gear - don’t waste your money on things that won’t up your experience.

 

Don’t let audiophile snobs fool you into buying hi-res stuff (24 bit/96 kHz), the advantage is purely technical. Files are more expensive and not worth the extra money.

 

Buy god speakers, it’s worth the investment.

 

Buy good cables but you don’t need audiophile level. That shit can be crazy expensive and you won’t hear any difference compared to medium price, high quality equipment. Trust me. I’ve poured more money into cables and connectors than I’d like to admit, so I’ve learned the hard way.

[Jurassic Shark 12,902]

I agree about averything in that post!

[Marian Schedenig 8,900]

10 hours ago, Jurassic Shark said:

And a decent DAC, either stand-alone or part of your amp or player.

 

And good headphones.

 

9 hours ago, rough cut said:

Buy good cables but you don’t need audiophile level. That shit can be crazy expensive and you won’t hear any difference compared to medium price, high quality equipment. Trust me. I’ve poured more money into cables and connectors than I’d like to admit, so I’ve learned the hard way.

 

For the digital part of the signal chain, a "normal" cable should suffice. Compressed streams (e.g. DTS) are a good way to check if the cable is alright - if you have transmission errors, a frame of the stream will be broken, and the audio will drop out until the next frame. If you don't get dropouts, the cable is alright. If you do get them, it's a bad cable, or it's too long for its shielding (I have some occasional issues with my 7m (?) coax connection to the other room, but not frequent enough to bother me into getting a better cable). So if your sources are hooked up digitally to your amp, and your headphones connect directly to the amp as well, the only cables you have to worry about are the speaker cables.

[Jurassic Shark 12,902]

13 minutes ago, Marian Schedenig said:

And good headphones.

 

12 hours ago, Manakin Skywalker said:

 

All you really need is any decent pair of modern headphones/speakers. There's not really anything else you need in terms of equipment.

 

[The Train Station 8,601]

Speakers won't work without an amplifier.

[Marian Schedenig 8,900]

11 minutes ago, Jurassic Shark said:

 

12 hours ago, Manakin Skywalker said:

 

All you really need is any decent pair of modern headphones/speakers. There's not really anything else you need in terms of equipment.

 

For some reason, my brain saw only "speakers".

 

I'm still perfectly happy with my AKG K701. Except that the pillows are so worn down they don't really pillow (as a verb) anymore. Apparently, replacements are available, for something like €20 per ear, while I could get the K702 for considerably less than I paid for my K701 back in the day, so I'm torn between not wanting to spend a (still) lot of money on new, probably slightly better headphones* and spending less (but much more than seems proper) to make my perfectly fine old ones comfy again.

 

*) Supposedly the sound is even smoother, and they have a detachable cable, which would be useful considering how often I yank mine to the floor.

 

[Jurassic Shark 12,902]

5 minutes ago, Marian Schedenig said:

Apparently, replacements are available, for something like €20 per ear, while I could get the K702 for considerably less than I paid for my K701 back in the day, so I'm torn between not wanting to spend a (still) lot of money on new, probably slightly better headphones* and spending less (but much more than seems proper) to make my perfectly fine old ones comfy again.

 

Often, you can find cheap third-party replacement cushions on eBay.

[Marian Schedenig 8,900]

58 minutes ago, Jurassic Shark said:

Often, you can find cheap third-party replacement cushions on eBay.

 

Oh yes, I found some others, I think they were €30 for both ears (including shipping). Still more than I fancy paying for a pair of pillows.

[Jurassic Shark 12,902]

I guess it could be worth it if the quality of the cushions is good, and you get to prolong the lifetime of your good headphones.

[Disco Stu 15,501]

Regarding high res digital audio, well, there's a sucker born every minute

[Kühni 484]

Query to the Gallery of Knowledgeables:

 

I've ripped the new Interstellar release to FLAC and out of curiosity put the files through two different pieces of audiochecking software.

 

1) Audiochecker 2.0 (screenshot excerpt):

 

2) Lossless Audiochecker:

 

Lossless Audio Checker 2.0.7 logfile from Monday 14 December 2020 02:19:46 PM

 

01 Dreaming of the Crash.flac
Result: Upsampled

02 Cornfield Chase.flac
Result: Upsampled

03 Dust.flac
Result: Upsampled

04 Day One.flac
Result: Upsampled

05 Stay.flac
Result: Clean

06 Message from Home.flac
Result: Upsampled

07 The Wormhole.flac
Result: Upsampled

08 Mountains.flac
Result: Clean

09 Afraid of Time.flac
Result: Upsampled

10 A Place Among the Stars.flac
Result: Upsampled

11 Running Out.flac
Result: Upsampled

12 I'm Going Home.flac
Result: Upsampled

13 Coward.flac
Result: Upsampled

14 Detach.flac
Result: Clean

15 S.T.A.Y.flac
Result: Upsampled

16 Where We're Going.flac
Result: Upsampled

01 First Step.flac
Result: Upsampled

02 Flying Drone.flac
Result: Upsampled

03 Atmospheric Entry.flac
Result: Upsampled

04 No Need to Come Back.flac
Result: Upsampled

05 Imperfect Lock.flac
Result: Upsampled

06 No Time for Caution.flac
Result: Clean

07 What Happens Now-.flac
Result: Upsampled

08 Who's They-.flac
Result: Upsampled

09 Murph.flac
Result: Upsampled

10 Organ Variation.flac
Result: Upsampled

11 Tick-Tock.flac
Result: Clean

12 Day One (Original Demo).flac
Result: Upsampled

13 Day One Dark.flac
Result: Clean

14 Do Not Go Gentle Into That Good Night.flac
Result: Upsampled

 

I've tried reading and understanding previous explanations, but I'm still having problems interpreting these results. The CDs were ripped with Exact Audio Copy on my eight year-old laptop.

 

Leaving aside how the programs and their algorithms work, Audiochecker 2.0 and Lossless Audio Checker are giving different results. The majority of ripped files are CD audio quality, but let's look at one example that A2.0 identified as an MPEG file, "Organ Variation". It's spectrogram looks as follows:

 

 

It's mainly a low profile, so is the program "misinterpreting" the file, which actually is completely fine? 

 

Bottom line: is it possible that for any reasons (old hardware, not-correct settings in the softwares used [Lossless Audio Checker has no settings to change, however]) the results include both false positives and/or false negatives? Upsampling doesn't change anything in the actual hearable quality of a track, or is that where (among others) where proof exists to differentiate between a CD quality source and, say, a converted mp3 source file?

 

And is there actually a difference between files ripped from a CD and those that, e. g. on Qobuz, are provided digitally? My apologies is what I'm asking/writing about is unclear, I unfortunately do not all-encompassingly know about this topic and its finer, technical aspects.

 

Ta!

[The Train Station 8,601]

Too many people seem to think audiophile quality is about some abstract whatchumacallit about ultra high frequencies or something. Where the hell did that come from? Nobody can hear that shit, so don't worry about it. What's important is dynamics and whether the audio is accurate and ensuring there's headroom so those peaks don't clip/distort.

[Kühni 484]

24 minutes ago, The Big Man said:

Nobody can hear that shit, so don't worry about it. What's important is dynamics and whether the audio is accurate and ensuring there's headroom so those peaks don't clip/distort.

 

Thanks for your reply, TBM. So I take it that an mp3/mp4 whose spectogram appears unclipped file tops a clipped/brickwalled lossless file? Really?

 

Another example: the spectograms for "Mountains" look, to me, the same between the AAC file I use for listening on laptop/iPod and the FLAC for hi-res storage, notwithstanding the filtering out of frequencies over 20 kHz for the AAC file.

 

Is it there that (all) the audible differences lay, if they actually do exist at those settings? (And the line at 16 kHz is an artifact stemming from the hardware, correct?)

 

 

 

I recently made a hearing test at the doctor's in a quiet room. My ears peace out at just over 14 kHz (but at elevated volume). Which, for all intents and purposes, is what it should be at my age and all things considered. Which means I would not be able to hear either the 16 kHz line nor notice the filtered-out high frequencies... 

[1977 1,745]

I only use lossless files (ALAC in my case) for archival purposes. All my non-CD listening is done in AAC or MP3 format.

 

In cases where I need to edit lossy files, I save them into AIFF format to avoid re-encoding and further quality loss.

[The Train Station 8,601]

46 minutes ago, Kühni said:

 

Thanks for your reply, TBM. So I take it that an mp3/mp4 whose spectogram appears unclipped file tops a clipped/brickwalled lossless file? Really?

 

Another example: the spectograms for "Mountains" look, to me, the same between the AAC file I use for listening on laptop/iPod and the FLAC for hi-res storage, notwithstanding the filtering out of frequencies over 20 kHz for the AAC file.

 

Is it there that (all) the audible differences lay, if they actually do exist at those settings? (And the line at 16 kHz is an artifact stemming from the hardware, correct?)

 

 

 

I recently made a hearing test at the doctor's in a quiet room. My ears peace out at just over 14 kHz (but at elevated volume). Which, for all intents and purposes, is what it should be at my age and all things considered. Which means I would not be able to hear either the 16 kHz line nor notice the filtered-out high frequencies... 

 

Those high frequencies aren't as much a big deal as you might think. The most important frequency range is about 1khz anyway, and the higher you go, the more it's there to merely support that range. 

 

As for your first question, yes. There are hi-rez files sold on HD Tracks that are brickwalled and sound like poop. An mp3 made from a more dynamic master would sound better.

[publicist 4,647]

There is another thing to consider, which is the codec. Codecs in the early 2000's were often dodgy and have been updated to quite sophisticated tools in the last 15 years. So if you nowadays use a good Lame or Apple codec (preferably at a bitrate higher than 224 k/bit), you will hardly note the quality loss.

 

Many so-called audiophiles are a joke - if you'd do the Pepsi test with them they would fail abysmally. You need really good gear and surroundings to be able to hear where a 320k file of i. e. a recent release like War of the Worlds loses fidelity compared to cd resolution. 

[Kühni 484]

Thank you very much for your replies, gentlemen. I'm happy I won't be sweating it overly much in the future when it comes to these things!

 

(Of course it's the codecs...codices...it's always them. We lost at Stalingrad because of faulty codecses, preciousss...)

[Marian Schedenig 8,900]

The thing is, compression codecs don't just filter out frequencies. I've never bothered to learn the details; I suppose they do that as well, but they also affect other characteristics by filtering out whatever "information" is deemed a) less relevant and b) expensive in terms of bandwidth. That gurgling water-like sound (first in percussive instruments, and with increased compression for other parts as well) that was so typical of standard bitrate MP3s in the earlier days (and which I expect you still get if you crank up the compression too much) certainly isn't the result of just dropping some parts of the overall frequency spectrum.

[thx99 1,796]

On 12/20/2020 at 6:20 AM, Kühni said:

(And the line at 16 kHz is an artifact stemming from the hardware, correct?)

 

 

That relatively constant tone heard/seen in these spectrograms at just below 16 kHz likely has to do with electromagnetic interference induced by the line frequency of a television or video monitor that was in proximity with the recording/re-recording equipment:

• NTSC: 525 lines * 29.97 fps = 15.734 kHz
• PAL: 625 lines * 25 fps = 15.625 kHz

[Datameister 2,241]

On 12/20/2020 at 7:35 AM, Marian Schedenig said:

The thing is, compression codecs don't just filter out frequencies. I've never bothered to learn the details; I suppose they do that as well, but they also affect other characteristics by filtering out whatever "information" is deemed a) less relevant and b) expensive in terms of bandwidth. That gurgling water-like sound (first in percussive instruments, and with increased compression for other parts as well) that was so typical of standard bitrate MP3s in the earlier days (and which I expect you still get if you crank up the compression too much) certainly isn't the result of just dropping some parts of the overall frequency spectrum.

 

Thissssssssss. The abrupt loss of high-frequency information can be a great visual clue on a spectrogram, but in terms of listening, those artifacts in the more audible frequency ranges are the potential problem. I know that in blind tests, I fail to distinguish between lossless originals and high-bitrate mp3 or AAC compression, even on high-end equipment. As the bitrate drops, the difference starts to become more and more obvious.

 

As @JTWfan77 said, lossless quality is great for archival purposes, and it's especially important if you're going to be editing the material at some point and then transcoding with lossy compression afterward. Re-compressing the same audio multiple times is a really easy avenue to awful sound quality. So I still strongly prefer lossless purchases, but my everyday listening is all with high-bitrate lossy transcodes.

[Jay 39,199]

4 minutes ago, Datameister said:

I still strongly prefer lossless purchases, but my everyday listening is all with high-bitrate lossy transcodes.

 

Same