Optimal Mp3 Bitrate
Jun 30, 2006 at 3:26 AM Post #31 of 49
Quote:

Originally Posted by Xyrium
Using EAC, a Flac encoded file was 22MB, while the 320kbps Lame MP3 was only 8MB. So, I guess I'll continue to use Lame since at age 36, my hearing is only being reduced to a frequency range of 500Hz-15kHz probably, with a sensitivity of about -65dB.
wink.gif



HAHA..... This is a very good point. My good friend is going into audiology and I am quite sure she couldn't agree with you more!
tongue.gif
 
Jun 30, 2006 at 6:37 AM Post #32 of 49
Quote:

Originally Posted by lordgibbness
The best top-of-the-range LAME setting I would suggest is "-V 0 --vbr-new" as quality wise it is on a par with CBR320, but will save you a reasonable amount of space.


On that chart (behind link), where would be the point (in generally), you're not able clearly to notice/hear the quality difference (quality 1-10/setting) [let's forget the "age parameter" now]?

jiitee
 
Jun 30, 2006 at 7:52 AM Post #33 of 49
Quote:

Originally Posted by jiiteepee
On that chart (behind link), where would be the point (in generally), you're not able clearly to notice/hear the quality difference (quality 1-10/setting) [let's forget the "age parameter" now]?

jiitee



-V 2, (or preset standard) is widely seen to be transparent encoding.
 
Jul 1, 2006 at 12:39 PM Post #34 of 49
Quote:

Originally Posted by krmathis
-V 2 is transparent for me.
At levels below this I am sometimes able to pick the lossy one in an ABX test.



I like how you're the only one who mentioned "abx."
 
Jul 1, 2006 at 1:37 PM Post #35 of 49
Quote:

Originally Posted by jiiteepee
On that chart (behind link), where would be the point (in generally), you're not able clearly to notice/hear the quality difference (quality 1-10/setting)


This is an impossible question to answer on an individual basis, because everyone's hearing is different. The best way to answer is to speak in generalities. As mentioned above, -V 2 should be transparent for the overwhelming majority of people--perhaps 99%. But higher settings will still be transparent to most people.

The best way to find out what is good for you is to do this:

1. Start by encoding a file to mp3 with one of the higher numbers, like -V 7 or even -V 9.

2. Use Foobar's ABX comparator to compare the mp3 to the original file. (Decide in advance how many trials you are going to do. It should be at least 10). The ABX comparator will report the likelihood that you were guessing.

3. If the likelihood that you are guessing is more than 5%, stop. You've found a setting that is transparent to you.

4. If the likelihood that you are guessing is less than 5%, it is safe to conclude that you can hear a difference between that level of encoding and the original. If so, then move to the next -V level and repeat the process.

The nice thing about going through this ABX process is that it allows you to generate the smallest files that are truly transparent to you. For example, I *thought* that I needed files to be encoded with at least -V 2 to be transparent. When I did the ABX test, I discovered that I really only need -V 5 for portable use. There are many people around here who will say that they use 320kbps CBR "just to be safe." That is really overkill; even the LAME developers don't advise using 320kbps CBR because the quality improvement between -V2 and 320kbps is marginal compared to the relatively large increase in file size. An ABX test can let you find the lowest bitrate that is transparent for you without wasting disk space.

Quote:

Using EAC, a Flac encoded file was 22MB, while the 320kbps Lame MP3 was only 8MB. So, I guess I'll continue to use Lame since at age 36, my hearing is only being reduced to a frequency range of 500Hz-15kHz probably, with a sensitivity of about -65dB.


If you are archiving, I would recommend using a lossless format like FLAC, not because it will offer any perceptible improvement to you over the 320kbps mp3s that you currently use, but because it offers more flexibility to moving to other formats or bitrates. IMHO, 320kbps is the worst of all worlds for archival purposes. It gives you files that are much bigger than files generated using the - V settings with little (if any) perceptible quality difference. However, if you want to switch from 320kbps to another format or bitrate, you need to do a second round of lossy encoding, which will result in a bigger quality loss than if you had encoded directly from the source to that other format or bitrate.
 
Jul 1, 2006 at 8:41 PM Post #36 of 49
Amen, Febs. That's one of the best posts I've seen around here in a while. I think a lot of people just assume they need -V 2 or higher for transparency. -V 6 is actually transparent to me for portable use.

I agree with you about lossless as well. It's not about percebtible quality improvements at all, it just makes transcoding to different formats much easier.
 
Jul 4, 2006 at 5:54 PM Post #37 of 49
Yeah, I'm now using the "insane" preset for LAME and it's probably more of a psychoacoustical difference than anything. Heck, if it makes me think I'm hearing a better rip, then so be it.
smily_headphones1.gif


--nspsytune -b320 -h -mj --nssafejoint --lowpass 20 --athtype 2 -X3

The only issue is that I need to use replaygain on many of my files due to clipping with ASIO on the 1212M.
 
Aug 19, 2007 at 6:54 PM Post #38 of 49
128 kbps mp3 always sounded horrible to me, so I used 192 or around 200 VBR as much as possible. That was sufficient when listening directly from my soundcard with HD595.

Now with my current setup, 192 kbps is not optimal anymore. I switched to 320 kbps.
After A/B testing I found a very slight but definite difference in refinement of details.
I also believe there's a slight difference in dynamics but that's a bit harder to test.
 
Aug 20, 2007 at 12:53 AM Post #41 of 49
I personally don't think I can hear a difference at 192, but I have the hard drive space, so why not go FLAC? I have some low quality recordings in 192, and I have some remnants from my iTunes days at 128, but I don't like to go that low. 320 is my ripping speed now for most recordings, just because I like the safety factor.

What made a difference for me was the upgrade of my audio card and the switch to using ASIO in lieu of kernal streaming.
 
Aug 20, 2007 at 2:39 AM Post #42 of 49
I use LAME -V4 in my DAP for most of the stuff, and -V0 if I need HQ. I use FLAC -6 at home.
 
Aug 20, 2007 at 4:27 AM Post #43 of 49
When I had very limited portable player space and was traveling quite a bit, I wanted to maximize compression for use on airplanes. To determine what my situation was in that environment, I used my laptop with Echo Indigo sound card and ER4S to do ABX testing in an airplane. In that environment V6 worked just fine for me and the V6 files were 40% the size of the V2 files. I always think I can hear more differences than I can demonstrate in an ABX test.

BTW as a baseline, I cannot reliably tell V2 from CD when ABXing at home, even though I had previously convinced myself that i could. YMMV.

I believe in placebo effect, which is to say that someone's total context/expectation can influence their enjoyment as they perceive the music in that context. So whether you can pick out lossless in ABX, if it makes you happy and you have the space, go for it. My little airplane experiment convinced me to squeeze more of my music into my little old 512 MB iAudio, which I still think has a great audio output.

Alan
 
Aug 20, 2007 at 5:03 AM Post #44 of 49
I can't tell any difference between 192kbs and a wave file but I rip to 256kbs just to make sure I don't hear a difference.
smily_headphones1.gif
I'm not wasting my HDD space on flac even though I have used flac in the past. If I can't hear the difference, and that is on half decent equipment too, why use flac?
 
Aug 20, 2007 at 9:10 AM Post #45 of 49
The best LAME 3.97 Final settings are, using lame.exe CLI:
best quality (midrange): --cbr -b 320 -q 0
Using higher HP filter frequency is stupid in this case because you waste many bitsa to encode highest frequencies (mostly inexistent or inaudible for many) sacrificing midrange quality and warmth. The same with Stereo against Joint Stereo - less bits for other purposes.
If you don't care about warmth, go this:
--vbr V0 -q 0 -m s
In this case you will get the best soundstage reproduction, to sacrifice it a bit and save ~10~80kb/s, discard "-m s"
The above setting makes more sense than 256kb/s for LAME.
For those unfamiliar with LAME CLI, go --preset-insane and --preset-extreme respectively.
 

Users who are viewing this thread

Back
Top