IPB

Welcome Guest ( Log In | Register )

2 Pages V   1 2 >  
Reply to this topicStart new topic
Lossless vs 320kbps vs 256kbps, Moderation: no actual ABX results to see here.
agatha1
post Nov 17 2011, 12:22
Post #1





Group: Members
Posts: 7
Joined: 17-November 11
Member No.: 95228



Hello everyone and thanks for making this forum what it is, I think the best audio forum in the world.

This is my first post here and I'm a little bit concerned about my ABX blind test results. I currently have over 6 TB of music, mostly FLAC and mp3 at 320kbps, but I also have some rips at 192 or below.

Now I want to standardize my whole collection and see if I can save a lot of Hard disk space or if I need to get better rips for part of my collection, etc. So, after some very interesting reading here, I decided to perform my own set of ABX blind tests, downloading foobar2000 and the ABX plugin.

Of course for some simple songs I can't tell any difference even between FLAC and 128 kbps, but I want a standard for my whole collection, including a lot of classical and progressive rock, so I need to find a bitrate that ensures me that I will ALWAYS have the best quality I can hear.

I tried the test first with Camarillo Brillo (by Frank Zappa), which I think it's a very good song for this test, because at the very beginning, after only 4 seconds, there is a kind of rattlesnake (sorry, I don't know the name of the instrument), which sounds definitely different at every bitrate. I encourage everybody to use this song for testing, as it makes it very quick.

Now, my results:
FLAC vs 256: I can always tell the difference, 10 out ot 10 times.
but here's the strange thing:
FLAC vs 320: Not clear
320 vs 256: Not clear.

¿Does this make any sense to you?
And if so, ¿Which is the bitrate I should stick to for my whole collection?

Thank you in advance for any help.

Regards from Spain,
Agatha.
Go to the top of the page
+Quote Post
db1989
post Nov 17 2011, 12:29
Post #2





Group: Super Moderator
Posts: 5275
Joined: 23-June 06
Member No.: 32180



1. Moved to Listening Tests
2. Post test logs, please
3. Using one song, and especially one with a specific type of artefact, is not an accurate way to predict the performance of a particular format/setting on your entire library. Test various tracks, genres, etc.
Go to the top of the page
+Quote Post
agatha1
post Nov 17 2011, 12:53
Post #3





Group: Members
Posts: 7
Joined: 17-November 11
Member No.: 95228



Hi db1989,

1.- Thank you for moving my post to the correct sub-forum, and sorry for my mistake.
2.- ¿How can I do that? OK, I'll read more and try to find out myself, but thank you if you can help me here.
3.- I guess you're right, but also it makes sense to me that if the 'rattlesnake' in Camarillo Brillo sounds sooo different at 256 from its original FLAC, that means that 256 is no good to me ¿Am I wrong?

I don't have very good hearing, and I know that 192 kbps is more than enough for probably more than 80% of my entire collection, but if I ever decide to lower my whole collection bitrates, first I want to be sure that I won't be missing *anything* that I can actually hear, so I guess 320 is my minimum (?)

Also, the main difference in hard disk saving goes when downgrading from FLAC to mp3, so I think I will not test 256 or lower any more. Instead, I will confront FLAC vs 320 with other songs, to see If I can *ever* tell any difference, since I could not find any with my Camarillo Brillo test.

Thanks again and keep up with the good work!
Go to the top of the page
+Quote Post
Jillian
post Nov 17 2011, 14:57
Post #4





Group: Members
Posts: 66
Joined: 29-April 06
From: Thailand
Member No.: 30166



I think the answer is very clear, if you're happy with 320kbps and not with 256kbps then you should stick with 320kbps. But I kinda like variable bitrate over constant, or move to other codecs if possible or preferred. I also wonder if you're properly test it, uncontrolled test can lead to bias result.
Go to the top of the page
+Quote Post
shadowking
post Nov 17 2011, 14:58
Post #5





Group: Members
Posts: 1527
Joined: 31-January 04
Member No.: 11664



IMO stay around 224k and not more as there is extra safety margin over 192 without using too high bitrate. The thing is there is still some music that is not transparent at 320k (rare). I think at 224k even most badly behaved stuff sounds okay or close to original.



--------------------
Wavpack -b450s0.7
Go to the top of the page
+Quote Post
mixminus1
post Nov 17 2011, 15:10
Post #6





Group: Members
Posts: 688
Joined: 23-February 05
Member No.: 20097



@OP: What MP3 encoder, and what settings were used?


--------------------
"Not sure what the question is, but the answer is probably no."
Go to the top of the page
+Quote Post
Northpack
post Nov 17 2011, 15:11
Post #7





Group: Members
Posts: 455
Joined: 16-December 01
Member No.: 664



You should compare 320kbps MP3 with variable bitrate, say -V3 and -V5. You can save a lot of space by choosing VBR over CBR.
Go to the top of the page
+Quote Post
agatha1
post Nov 17 2011, 15:43
Post #8





Group: Members
Posts: 7
Joined: 17-November 11
Member No.: 95228



QUOTE (mixminus1 @ Nov 17 2011, 15:10) *
@OP: What MP3 encoder, and what settings were used?


I'm using Easy CD-DA extractor version 15.3.2., which includes Lame, version 3.99. The only settings are CBR, 320 kbps, Auto (for Stereo/Mono/Joint Stereo), Highest (for Quality: Highest-High-Normal-Fast-Very Fast) and Auto (for samplerate), with lowpass and highpass both disabled.

I've tested my hearing at noiseaddicts and I cannot hear anything over 15kHz, so I gues I could save some bytes there with the filters. huh.gif
Go to the top of the page
+Quote Post
pdq
post Nov 17 2011, 16:51
Post #9





Group: Members
Posts: 3405
Joined: 1-September 05
From: SE Pennsylvania
Member No.: 24233



Yes, you should never disable highpass as quality could suffer.

Edit: I mean lowpass!

Edit2: You should also switch to a decent ripper.


This post has been edited by pdq: Nov 17 2011, 17:12
Go to the top of the page
+Quote Post
agatha1
post Nov 17 2011, 20:13
Post #10





Group: Members
Posts: 7
Joined: 17-November 11
Member No.: 95228



QUOTE (pdq @ Nov 17 2011, 16:51) *
Yes, you should never disable highpass as quality could suffer.

Edit: I mean lowpass!

Edit2: You should also switch to a decent ripper.


Can you please recommend a better ripper for downgrading my FLACs to top quality mp3?
Honestly, I thought all of them were the same, as long as they're using the same codec (Lame 3.99 not good?).
Thank you in advance.
Go to the top of the page
+Quote Post
pdq
post Nov 17 2011, 20:30
Post #11





Group: Members
Posts: 3405
Joined: 1-September 05
From: SE Pennsylvania
Member No.: 24233



Converting from one compressed format to another is not ripping. Ripping is the process of reading the audio data off of a CD.

The best paid (non-free) ripper is dBpoweramp. There are several excellent free rippers, such as EAC. IMHO, dBpoweramp has some advantages over the free rippers.
Go to the top of the page
+Quote Post
kwanbis
post Nov 17 2011, 22:10
Post #12





Group: Developer (Donating)
Posts: 2362
Joined: 28-June 02
From: Argentina
Member No.: 2425



QUOTE (agatha1 @ Nov 17 2011, 19:13) *
Can you please recommend a better ripper for downgrading my FLACs to top quality mp3?

Foobar2000 would be the best option.


--------------------
MAREO: http://www.webearce.com.ar
Go to the top of the page
+Quote Post
Soap
post Nov 17 2011, 22:28
Post #13





Group: Members
Posts: 1015
Joined: 19-November 06
Member No.: 37767



Let's fix the terminology here. I think it is leading to confusion.

In popular parlance "ripping" is the act of extracting data from the optical disc.

"Encoding" is converting that ripped audio to a compressed (FLAC / MP3 / etc) format.

"Transcoding" is converting from one compressed format to another. Decoding then Encoding.

You don't want a ripper to convert FLAC to MP3 - you want an encoder or method of transcoding.

EDIT: Oops, saw kwanbis' post reusing the confusing terms and didn't see that pdq had also posted a new reply making the same point.


This post has been edited by Soap: Nov 17 2011, 22:30


--------------------
Creature of habit.
Go to the top of the page
+Quote Post
astroidmist
post Nov 18 2011, 00:34
Post #14





Group: Members
Posts: 34
Joined: 10-January 11
Member No.: 87208



Hard drive space is always getting cheaper (except for recent hard drive factory floodings in Asia). I would go with FLAC since you may have guests over who can hear better than you. That way everyone is guaranteed to enjoy the best fidelity if it's there. And if you need lossy copies to fit on a portable player from this age then you can generate them from the FLACs. But even portable player storage is increasing as time goes on. And for that matter, used CD prices are pretty low these days too.


--------------------
opinion is not fact
Go to the top of the page
+Quote Post
agatha1
post Nov 18 2011, 01:34
Post #15





Group: Members
Posts: 7
Joined: 17-November 11
Member No.: 95228



Thank you all for your answers.
The thread is becoming increasingly more and more interesting for me!
Now I find my first original question makes less sense, at least at this time, and instead there's some new questions that arise before I can proceed any further.

1.- RIPPING AND LOSSLESS ENCODING ( From CD to FLAC).
I ripped my collection of original CDs ages ago, encoded them to FLAC and never worried about them any more. I don't like the Optical Disc-Laser Lens system for playing back anything, since I got a lot of them scratched, and also my room full with thousands of CDs and DVDs, so at some point I decided to rip them all and keep them stored in hard disks, both movies (as mkv) and audio (as FLAC).

I think I used foobar at that time, but here is my first new question: If all lossless formats are what they claim, i.e., lossless, ¿What is the difference if you use one ripper or the other, or one output format or the other (wav, ape or flac), aren't they all exactly the same?

Also some hundreds of FLACS I've downloaded directly from the internet, I only checked that they were really lossless and error free, then added them to my collection, in the belief that they were indistinguishable from my original CD rips. ¿Aren't they? After all, I'm supposed to be able to rebuild the original CD from any of these sources, so ¿What's the difference between them?

Not to mix things, I mean always at the same 16 bit, 44.1 kHz, or whatever, same as in the original source. I have also downloaded some 24 bit 192 kHz rips, but I never could tell any difference (though I know, that's just me).

2.- TRANSCODING (From FLAC to mp3)
I need to transcode part of my collection from FLAC to mp3. Maybe a 1TB hard disk that I can play inside the car or anywhere else where a FLAC player is not available. I know that even 192 kbps or a decent VBR would be acceptable while driving, but anyway I would like learn how do the transcoding right, just in case.

From my first blind tests, I was surprised that I could tell a 256 kpbs mp3 from the FLACs, at all times (even with Bob Dylan or Janis Ian!). I know my hearing is way below the average, but all songs sounded so different. However, I could never tell the difference between the FLAC and the 320 kps mp3. Also a lowpass filter as low as 15 kHz seems to make no difference to my ears!

So here is my second new question: ¿Why is it a good or a bad idea to enable/disable the highpass and lowpass filters in Lame? ¿Something to do with harmonics and resonance? I couldn't tell any difference up till now, ever, but I would just like to learn and know more. TIA.
Go to the top of the page
+Quote Post
kwanbis
post Nov 18 2011, 04:00
Post #16





Group: Developer (Donating)
Posts: 2362
Joined: 28-June 02
From: Argentina
Member No.: 2425



If all lossless formats are what they claim, i.e., lossless, ¿What is the difference if you use one ripper or the other, or one output format or the other (wav, ape or flac), aren't they all exactly the same?
The ripper just extracts the information from the physical CD, then the encoder creates the lossless file. So they are 2 different things, even if NORMALLY, the ripper runs the encoder all by itself. So, it is important the ripper, cause if the ripped tracks have errors, the lossless files, no matter what format you use, would also have errors. There are rippers that can handle errors well (EAC and dbpoweramp on windows), and some that don't. But once you have a good rip, any lossless encoder would produce a file that contains the same musical content, as other lossless encoders.

Also some hundreds of FLACS I've downloaded directly from the internet, I only checked that they were really lossless and error free, then added them to my collection, in the belief that they were indistinguishable from my original CD rips. ¿Aren't they? After all, I'm supposed to be able to rebuild the original CD from any of these sources, so ¿What's the difference between them?
Well, if the one who ripped them used a secure ripper, and then encoded with a lossless encoder, you should be fine, and no difference.

From my first blind tests, I was surprised that I could tell a 256 kpbs mp3 from the FLACs, at all times (even with Bob Dylan or Janis Ian!). I know my hearing is way below the average, but all songs sounded so different. However, I could never tell the difference between the FLAC and the 320 kps mp3. Also a lowpass filter as low as 15 kHz seems to make no difference to my ears!
Have you done a double blind test? Unless you have very special ears, is very doubtful you can distinguish them.

So here is my second new question: ¿Why is it a good or a bad idea to enable/disable the highpass and lowpass filters in Lame? ¿Something to do with harmonics and resonance? I couldn't tell any difference up till now, ever, but I would just like to learn and know more. TIA.
Just use lame -V 0 or -V 2 and let LAME decide.

This post has been edited by kwanbis: Nov 18 2011, 04:03


--------------------
MAREO: http://www.webearce.com.ar
Go to the top of the page
+Quote Post
pdq
post Nov 18 2011, 04:07
Post #17





Group: Members
Posts: 3405
Joined: 1-September 05
From: SE Pennsylvania
Member No.: 24233



All lossless formats are identical in quality. Where they differ is in such things as compressed file size, speed, features and compatibility.

But, you still don't get the ripping part. Ripping is potentially a LOSSY process! The wrong ripper can give you tracks with read errors and never even tell you that. If you ripped your CDs awhile ago then you probably didn't have the benefit of AccurateRip. I don't even know if foobar had secure ripping back then.

As for your being able to easily distinguish 256 cbr mp3 files from FLAC, that says to me that the ripper that you are using to encode to lame is probably buggy. There is absolutely no reason that virtually all of your 256 encodes shouldn't sound identical to the original.

Edit: kwanbis was quicker.

Edit2: The default lowpass filter settings in lame were selcted as the best compromise between including high frequensies that almost noone can hear, vs. instead using those bits to more accurately encode the lower frequencies that we all can hear. Don't mess with those settings unless you really know what you are doing. I for one could set a much lower lowpass since I can't hear anything above 12 kHz.

This post has been edited by pdq: Nov 18 2011, 04:17
Go to the top of the page
+Quote Post
Glenn Gundlach
post Nov 18 2011, 05:35
Post #18





Group: Members
Posts: 369
Joined: 19-April 08
From: LA
Member No.: 52914



I don't know what hard drives go for in Spain but in the US you could get 3TB for $120. If you stored your music in .FLAC you could have about 6000 CDs lossless and not worry about MP3 at all. Other than saving space, what good is MP3? All I hear about it is complaints.

Go to the top of the page
+Quote Post
db1989
post Nov 18 2011, 11:38
Post #19





Group: Super Moderator
Posts: 5275
Joined: 23-June 06
Member No.: 32180



QUOTE (Northpack @ Nov 17 2011, 14:11) *
You should compare 320kbps MP3 with variable bitrate, say -V3 and -V5. You can save a lot of space by choosing VBR over CBR.
I presume this is a short way of suggesting that the OP additionally compare VBR settings to lossless, rather than ABXing one lossy encode against another? The latter is illogical and of little, if any, use.
Go to the top of the page
+Quote Post
agatha1
post Nov 18 2011, 11:40
Post #20





Group: Members
Posts: 7
Joined: 17-November 11
Member No.: 95228



* For Glenn Guldlach and astroidmist:

I agree with both of you that hard disk space is not an issue anymore, and so I've decided to keep my collection of FLACs, untouched, whichever my final conclusion after these tests may be. Today I've got 3 x 2TB hard disks for music, the first one covering from 0 to L, the second one from M to Z, and the third one for classical and Spanish music, and I will not downgrade them, ever.

However, I'm still interested in learning how transcoding really works, especially with Lame, and which parameters I should use. Also, I have recently purchased a 'small' (1 TB) portable hard disk for car listening, and my car player won't play lossless, so I need to go mp3 (always keeping my flac originals at home).

* For kwanbis and pdq:

I don't mean to have *good* hearing because the <320 kpbs mp3 sounded so bad to me. In fact, I think that it could be my *bad* hearing what makes that happen (Could that be?).

As I said, I can't hear anything above the 15 kHz tone, and I mean nothing at all, and also I have been diagnosed a hearing loss at around 3 kHz in my left ear, and that is medical, I have the curves showing the drop in decibels and so on. As I see it, we all have a 'special', or 'personal' hearing at least, and there's not two of us who can hear exactly the same thing when the same song is being played.

Having said that, I understand that over 90% of the people should be happy with the same set of Lame parameters, and there's a lot of placebo effect when so many claim to hear things that they actually don't. I try my best not to do that, honestly, and I'm ready to admit that something I'm doing wrong with my transcoding, but I'm still testing changing a lot of parameters here and there and every ABX test comes to show that my ears are quite sensitive to any lowering of the bitrate, while there's a null reaction to the lowpass filters, no matter how low I set them.

I'll keep on testing and reporting when I find something new.
Thank you all for your great feedback!

Go to the top of the page
+Quote Post
pdq
post Nov 18 2011, 12:54
Post #21





Group: Members
Posts: 3405
Joined: 1-September 05
From: SE Pennsylvania
Member No.: 24233



Indeed it is possible for a defect in hearing to make one more sensitive to certain kinds of artifacts, though I'm not convinced that is what is happening here.

Are you performing proper double blind ABX tests?

Have you checked the sizes of your files to make sure the 256 kbps files are really 256 kbps?
Go to the top of the page
+Quote Post
db1989
post Nov 18 2011, 13:01
Post #22





Group: Super Moderator
Posts: 5275
Joined: 23-June 06
Member No.: 32180



And it would still be helpful to post the logs of these ABX tests, as I said above. Perhaps also take a clip (30 s or less) of the relevant files, compress it using the same settings, and post the lossless and lossy versions to the Uploads subforum.
Go to the top of the page
+Quote Post
kwanbis
post Nov 18 2011, 17:22
Post #23





Group: Developer (Donating)
Posts: 2362
Joined: 28-June 02
From: Argentina
Member No.: 2425



QUOTE (agatha1 @ Nov 18 2011, 10:40) *
However, I'm still interested in learning how transcoding really works, especially with Lame, and which parameters I should use. Also, I have recently purchased a 'small' (1 TB) portable hard disk for car listening, and my car player won't play lossless, so I need to go mp3 (always keeping my flac originals at home).

Since FLAC is an exact copy of the original master, just compressed, you can feed the FLAC to any encoder, and it would be like if you have taken the CD an encoded directly to that format.

So, if you take your FLACs, and encode to MP3, you would be fine.

I highly suggest you use LAME -V 0 or -V 2 for that.

You can use foobar2000, which is really simple.


--------------------
MAREO: http://www.webearce.com.ar
Go to the top of the page
+Quote Post
Remedial Sound
post Nov 18 2011, 17:50
Post #24





Group: Members
Posts: 508
Joined: 5-January 06
From: Dublin
Member No.: 26898



QUOTE (pdq @ Nov 18 2011, 06:54) *
Indeed it is possible for a defect in hearing to make one more sensitive to certain kinds of artifacts, though I'm not convinced that is what is happening here.

Are you performing proper double blind ABX tests?

Have you checked the sizes of your files to make sure the 256 kbps files are really 256 kbps?


There might also be a subtle loudness difference between the lossy and lossless versions being tested that's tipping off which is which. I haven't ABX'd in a long while but IIRC foobar's implementation levels the loudness of the audio samples to keep this source of error in check.

Knowing what we know of lame and transparency, it's doubtful (but not impossible, I suppose) that someone can consistently ABX 256 kbps across a broad range of music. Test logs would be good.
Go to the top of the page
+Quote Post
greynol
post Nov 18 2011, 19:02
Post #25





Group: Super Moderator
Posts: 10000
Joined: 1-April 04
From: San Francisco
Member No.: 13167



QUOTE (agatha1 @ Nov 18 2011, 02:40) *
I'm ready to admit that something I'm doing wrong with my transcoding

If you're transcoding lossy to lossy, then this is definitely something you're doing wrong.


--------------------
I should publish a list of forum idiots.
Go to the top of the page
+Quote Post

2 Pages V   1 2 >
Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 18th September 2014 - 14:33