Bit Rates and Sound Quality and Spectrograms

[mrcd]

What's the opinion of the authors of this site and others regarding bit rates? Some folks say you can't (and the human ear can't) hear any difference between a 128k , 320k, or a FLAC file after it's been ripped. Because of that they says there no reason to rip to higher bit rates. What of spectrograms that visually display audio patterns? Some say they are not reliable.

Also one person claimed nothing is lost in size or quality by converting from mp3 to aac and back to mp3.

[Spoon]

Re: Bit Rates and Sound Quality and Spectrograms

You listen to audio with your ears, not eyes, that is why spectrograms are no use.

[mville]

Re: Bit Rates and Sound Quality and Spectrograms

Some folks say you can't (and the human ear can't) hear any difference between a 128k , 320k, or a FLAC file after it's been ripped. Because of that they says there no reason to rip to higher bit rates.

Some folks can hear the difference, some folks can't and some folks don't care. It does depend on several factors, like playback hardware, listening environment, age etc. In the end it becomes a personal choice, like choosing to buy and listen to hi-res downloaded tracks, standard CDs or streaming from Spotify.

Also one person claimed nothing is lost in size or quality by converting from mp3 to aac and back to mp3.

Well, it may be that this person could not hear any difference, but this statement is untrue. mp3 to aac is lossy and aac to mp3 is lossy and so quality is affected.

[Oggy]

Re: Bit Rates and Sound Quality and Spectrograms

What's the opinion of the authors of this site and others regarding bit rates? Some folks say you can't (and the human ear can't) hear any difference between a 128k , 320k, or a FLAC file after it's been ripped. Because of that they says there no reason to rip to higher bit rates. What of spectrograms that visually display audio patterns? Some say they are not reliable.

Also one person claimed nothing is lost in size or quality by converting from mp3 to aac and back to mp3.

As well as using dBpoweramp, I also have an iPod and now rip at 320k. Even in a noisy environment at work, people commented on the (lack of) sound quality at 128k, preferring the same track at 256k. Spoon is absolutely right, you listen to music with your ears. The only reason these days to rip in a compressed format, is for space, and storage is now very cheap. You get a lot of songs on a 160GB iPod at 320k, and for lossless, a 2-3TB HDD, is more affordable than they have ever been.

An amplifier and speakers can only ever reproduce what is fed into them, and on a better system you will soon regret a low bitrate rip. Garbage in, garbage out! You soon get to appreciate a Secure or AccurateRip from CD Ripper - from a lossless file.

[garym]

Re: Bit Rates and Sound Quality and Spectrograms

In numerous controlled experiments (double blind tests), it becomes rare to find test subjects who can detect the difference between a lossless version and a lossy version (e.g., mp3), once an mp3 gets to *about* 190 bitrate. Obviously, there are known "killer samples" where it is relatively easy to detect lossy, even at 320kbs. Things like certain harpsichord, etc. But for real music (not just killer samples), 99.9% of the time it gets very hard to detect lossless from lossy. Most "golden ear audiophiles" that claim that they can easily distinguish lossless from 320kbs mp3 refuse to actually test this in a rigorous way (and come up with all kinds of wacko nonsense as to why ABX and doubleblind tests don't work, with most of the statements indicating they don't even actually understand how such tests work). I've done ABX tests myself on a variety of music, in a pretty good system. I'm never able to distinguish anything from lossless if above 190kbs (and created with a good mp3 encoder, ,such as lame). However, I have been able to more reliably detect mp3s from lossless at 128kbs or 160kbs *in some, but not all cases". You should try for yourself. Install foobar2000 and the ABX Comparator component. Then take a few different lossless files and create lossy versions. Create 320kbs, 190kbs, 128kbs, 96kbs, 64kbs, etc. Start testing with ABX and eventually with a low enough bitrate you'll start being able to reliably detect differences. The threshhold is different for everyone.

All this said, I choose to rip my CDs to lossless FLAC and I play these FLAC files in my home system. Space is cheap and small in size these days. Why not rip once and make it lossless. Then you're done no matter what. For my portables, I create a lossy mirror library of my files (convert from FLAC to mp3 at -V2 (~190kbs). For portable use, I'm sure something like -v5 would actually be fine too. The beauty of having the lossless orginals is I can always create a lossy mirror at smaller size (or using some new and improved lossy codec) in the future with no problem. (As it is a no-no to convert from lossy to lossy--this *may* not be audible, but it's not a good idea. You can test this too. create a mp3 from mp3, then do it again and again and again. Eventually, you'll create a lossy mp3 file that is distinguishable from the original mp3. But even here, it will likely take many many copies if one is using a high enough bit rate (e.g., 256kbs).

[Oggy]

Re: Bit Rates and Sound Quality and Spectrograms

In numerous controlled experiments (double blind tests), it becomes rare to find test subjects who can detect the difference between a lossless version and a lossy version (e.g., mp3), once an mp3 gets to *about* 190 bitrate. Obviously, there are known "killer samples" where it is relatively easy to detect lossy, even at 320kbs. Things like certain harpsichord, etc. But for real music (not just killer samples), 99.9% of the time it gets very hard to detect lossless from lossy. Most "golden ear audiophiles" that claim that they can easily distinguish lossless from 320kbs mp3 refuse to actually test this in a rigorous way (and come up with all kinds of wacko nonsense as to why ABX and doubleblind tests don't work, with most of the statements indicating they don't even actually understand how such tests work). I've done ABX tests myself on a variety of music, in a pretty good system. I'm never able to distinguish anything from lossless if above 190kbs (and created with a good mp3 encoder, ,such as lame). However, I have been able to more reliably detect mp3s from lossless at 128kbs or 160kbs *in some, but not all cases". You should try for yourself. Install foobar2000 and the ABX Comparator component. Then take a few different lossless files and create lossy versions. Create 320kbs, 190kbs, 128kbs, 96kbs, 64kbs, etc. Start testing with ABX and eventually with a low enough bitrate you'll start being able to reliably detect differences. The threshhold is different for everyone.

All this said, I choose to rip my CDs to lossless FLAC and I play these FLAC files in my home system. Space is cheap and small in size these days. Why not rip once and make it lossless. Then you're done no matter what. For my portables, I create a lossy mirror library of my files (convert from FLAC to mp3 at -V2 (~190kbs). For portable use, I'm sure something like -v5 would actually be fine too. The beauty of having the lossless orginals is I can always create a lossy mirror at smaller size (or using some new and improved lossy codec) in the future with no problem. (As it is a no-no to convert from lossy to lossy--this *may* not be audible, but it's not a good idea. You can test this too. create a mp3 from mp3, then do it again and again and again. Eventually, you'll create a lossy mp3 file that is distinguishable from the original mp3. But even here, it will likely take many many copies if one is using a high enough bit rate (e.g., 256kbs).

This is rather more scientific than my approach, and offers such good advice. I took an old Denon amp and Mission speakers into work, to help the working day pass better. Although not exactly top end audiophile equipment, it didn't sound bad.

With 64 and 128kbs mp3 source, after 7 hours of listening, headaches were a frequent occurrence, these were much reduced when 256 / 320k rates were used, and an audibly better sound was heard. I've never tried the 190kbs sample rate, but certainly 128 to 256 is marked - cymbals have attack and decay, and it is easier to follow bass lines.

The advice to rip to lossless and convert to whatever rate you are comfortable with, always keeping a master lossless copy, is excellent and most sensible. I've only just started playing with the Music Converter section of dBpoweramp, but it looks so flexible and will give you the above option with ease.

[mrcd]

Re: Bit Rates and Sound Quality and Spectrograms

There was a person who said on a forum that since you do listen with your ears and not depend on a graph it doesn't matter if you recode a 128k track to a 320k. In his mind it's a 320k now. Certainly if a track is a true to its bitrate visual tools must be able to offer some assistance otherwise how would you know a FLAC isn't really just a dressed up 128k?

[garym]

Re: Bit Rates and Sound Quality and Spectrograms

There was a person who said on a forum that since you do listen with your ears and not depend on a graph it doesn't matter if you recode a 128k track to a 320k. In his mind it's a 320k now. Certainly if a track is a true to its bitrate visual tools must be able to offer some assistance otherwise how would you know a FLAC isn't really just a dressed up 128k?

it doesn't matter going from 128 to 320, if one can't hear the difference between this recoded file and the original CD (lossless) in playback. And this could certainly be true in many cases. Regarding not knowing if FLAC is dressed up 128k. I'm sure some are. Heck there are Hi-res files sold as 24/192 that are actually originally sourced from low-bitrate mp3 files. Buyer beware!

[mrcd]

Re: Bit Rates and Sound Quality and Spectrograms

it doesn't matter going from 128 to 320, if one can't hear the difference between this recoded file and the original CD (lossless) in playback. And this could certainly be true in many cases. Regarding not knowing if FLAC is dressed up 128k. I'm sure some are. Heck there are Hi-res files sold as 24/192 that are actually originally sourced from low-bitrate mp3 files. Buyer beware!

Doesn't transcoding though from 128 to 320 just increase the file size only?

[garym]

Re: Bit Rates and Sound Quality and Spectrograms

Doesn't transcoding though from 128 to 320 just increase the file size only?

It does two things. It does increase the file size. But it *also* reencodes the file in a lossy manner, with a second generation of lossy encoding. The conversion to a different lossy format is not a LOSSLESS process.

[linkman]

Re: Bit Rates and Sound Quality and Spectrograms

My hearing is pretty good and I'm content with ~192kbps MP3 encoding. Most of my listening is through Logitech Z623 computer speakers (which are pretty good THX rated speakers) or moderately priced earbuds. I would probably notice very few differences between the 192kbps MP3 and a lossless version. I'm 50 years old and can still hear the horizontal oscillator on CRT based TV sets.

I notice a dramatic difference between 128kbps and 192kbps. I have a few of those lower bitrate songs in my collection and they sound terrible to me. I wish I could find better versions of them.

There's the real world and then there's near-perfect listening conditions.

[Oggy]

Re: Bit Rates and Sound Quality and Spectrograms

My hearing is pretty good and I'm content with ~192kbps MP3 encoding. Most of my listening is through Logitech Z623 computer speakers (which are pretty good THX rated speakers) or moderately priced earbuds. I would probably notice very few differences between the 192kbps MP3 and a lossless version. I'm 50 years old and can still hear the horizontal oscillator on CRT based TV sets.

I notice a dramatic difference between 128kbps and 192kbps. I have a few of those lower bitrate songs in my collection and they sound terrible to me. I wish I could find better versions of them.

There's the real world and then there's near-perfect listening conditions.

These are probably real world typical scenarios for the majority of users and possibly at a lossy level that is acceptable for many.

In the real world, you have got everything from budget phones / mp3 players being used, up to expensive DACs or streamers intended to compliment, or be the main source in top end Hi-Fi systems. What is acceptable on £10 earbuds, may not sound great through £10,000 speakers. These days a good music playing system doesn't seem to be a high priority for the vast majority.

The great thing with dBpoweramp, is that it caters for all expectation levels, so choose your poison carefully!

Used to own a pair of speakers with super tweeters, in isolation, they were practically inaudible. However, when playing music with just the super tweeter powered off, the effect was marked - honestly don't know what a spectrogram would show.