Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Lame 3.94a13 (Read 15377 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Lame 3.94a13

Updates: medium bitrate abr/cbr presets
new presets: portable/portable1/portable2/portable3.

I'd like your help on selecting the right portable preset.

Lame 3.94a13

Reply #1 –
win32 binaries at 'Other Stuff' Mirror 1.

Lame 3.94a13

Reply #2 –
Quote
win32 binaries at 'Other Stuff' Mirror 1.

Do you sleep? 

Lame 3.94a13

Reply #3 –
Actually, it's late afternoon in Britain now.

Who would sleep at such time in Easter?

Lame 3.94a13

Reply #4 –
Does this fix the problems you spoke of in alpha12??
r3mix zealot.

Lame 3.94a13

Reply #5 –
Yes, it should fix the cbr/abr problem

Lame 3.94a13

Reply #6 –
@gabriel: will there be a feature freeze (no updates from takehiro anymore) before fine-tweaking the codec??

testing new presets etc doesn't make sense if there are always updates to the psychoacoustics.

Lame 3.94a13

Reply #7 –
I do not know if there will be more major updates from Takehiro in the 3.94, but I do not think so.

But still, I need opinions on the current presets in order to know in which direction to go.

Lame 3.94a13

Reply #8 –
If LAME can be improved somehow, for a new version - LAME v4- I'd suggest:

- usage of MPC's psymodel: isn't it said to be the most accurate out there?
- no more n00b commandline support please....that should be removed IMO. Only clean
--alt-presets and the classic CBR modes.

Lame 3.94a13

Reply #9 –
What about --preset standard, any improvement? Is better that 3.90.2 or not yet?
MPC: --quality 10 --xlevel (v. 1.15s) (archive/transcoding)
MP3:  LAME 3.96.1 --preset standard (daily listening/portable)

Lame 3.94a13

Reply #10 –
I think that there should be a preset with an average bitrate of about 192 (like in older versions). Because a lot of us used preset-standard and now, its average bitrate is about 210. That's too much, I think. And the medium preset has only an average bitrate of 170.
Why do you have changed the bitrate of standard-preset?

Big_Berny

Lame 3.94a13

Reply #11 –
Quote
I'd like you help on selecting the right portable preset.

Only one thing about this is clear to me: Obviously the bitrate should be quite a bit lower than standard. (From what I've tested so far the results are 120 ... 130-140 ... 150 kbps).

I have no idea if there are clearly defined goals about
- quality
- speed

My thoughts about these points:

Probably at the targeted bitrates most pieces of music will contain positions where a trained ear can ABX differences. Artifacts that are even noticable without comparing to the original will occur regularly. So before starting to tune the lower bitrate presets there should be a discussion about what kind of artifacts should be avoided as much as possible and what would be tolerable if a decision must be made. For me e.g. pre-echo isn't that bad (talking about portable use - running, listeing while driving) but ringing (e.g. harsh "s"s) and warbeling/flanging ("boiling oil applause") are really annoying to me.

IMO the lower the bitrate the more important gets enccoding speed. If I want to put 10 hours of mp3 music on a CD-RW e.g. transcoded from my lossless/mpc collection on PC I don't want to wait 7 hours for encoding. No matter how good --preset portable will become quality-wise: If it encodes at ~ 1,5x speed (as 3.90.2 aps) here (Duron 750) I will prefer --alt preset e.g.140 (~2,5x speed) or gogo (7-9x speed at vbr settings for similar bitrate).

So please consider at least adding speed-focussed types of the lower bitrate presets (like --alt-preset fast ...).
-------------------------------
So far I encoded 2 songs with the portable presets. I haven't spent much time on it but I found 1 positon with obvious "warbel" and ringing "s" in --preset portable. pp1/2/3 and 3.90.2 --alt-preset 140 didn't have the problem. I can provide a sample if wanted.
Let's suppose that rain washes out a picnic. Who is feeling negative? The rain? Or YOU? What's causing the negative feeling? The rain or your reaction? - Anthony De Mello

Lame 3.94a13

Reply #12 –
Quote
I think that there should be a preset with an average bitrate of about 192 (like in older versions). Because a lot of us used preset-standard and now, its average bitrate is about 210. That's too much, I think. And the medium preset has only an average bitrate of 170.
Why do you have changed the bitrate of standard-preset?

Big_Berny

Well, the bitrate depends A LOT on what type of music you are encoding. I tested 3.90.2 vs an earlier alpha version of 3.94 (dont remember which), and on my type of music (pop/rock) the bitrate actually decreased with ~5 kbps on avarage over like 15 albums. But on individual albums and tracks the difference could be as much as 20-30 kbps, both higher and lower than 3.90.2.

Lame 3.94a13

Reply #13 –
Let's clarify all this:

Target bitrates are:
extreme: 240
standard: 200
medium: 165
portable: 130
streaming: 55 ???

There should be fast variants for all those, but as fast and regular vbr modes are quite different, I'd like to nearly finish regular vbr before trying fast vbr.
The standard bitrate really seems to be 200 ON AVERAGE.

Things that should not change much:
abr/cbr: should be ok now
vbr from medium to extreme: should only change a little

Things that need some quick input: portable presets. No need for deep test, but indications about which one to select would be very usefull.

Things that need to be done:
very low bitrate abr preset??
fast vbr presets


Current standard preset is better than 3.90.2 on some samples, but there are some important regressions on a few samples. There is need to work on those few samples.

I can not and should not tell you if the current standard is better or worst than the 3.90.2, as that would mean relying only on my ears, and I think that you do not want that.
Instead it is testers that should tell if it is better or no.

Lame 3.94a13

Reply #14 –
Quote
Things that need some quick input: portable presets. No need for deep test, but indications about which one to select would be very usefull.

OK. I say kick --preset portable. About the other p1/p2/p3 i'm not sure yet.
Let's suppose that rain washes out a picnic. Who is feeling negative? The rain? Or YOU? What's causing the negative feeling? The rain or your reaction? - Anthony De Mello

Lame 3.94a13

Reply #15 –
update: For testing I used
Omara portuondo - Quizas, quizas, quizas 0:24 - 0:30
Result:
Warbling + sharp "s" in
--preset portable
--preset portable2
--preset portable3
I can't tell which one is worst...

--preset portable1 sounds ok
3.90.2 --a-p 140 sounds ok
Let's suppose that rain washes out a picnic. Who is feeling negative? The rain? Or YOU? What's causing the negative feeling? The rain or your reaction? - Anthony De Mello

Lame 3.94a13

Reply #16 –
Just from some quick listening to the portable presets here's how I rate them:

Bruce Dickinson - Jerusalem (Live)

--preset portable: 159 kbps (Heavy flanging, not useable at all - the worst)
--preset portable1: 170kbps (Slight flanging, best of the 4)
--preset portable2: 149kbps (Heavy flanging, only slightly better than "portable" still unuseable)
--preset portable3: 152kbps (Easily better than portable & portable2, comparable quality to portable1)
--preset 130: 127kbps (Flanging though less annoying than portable & portable2)

Lame 3.94a13

Reply #17 –
I've tested all four on about 10 files so far.  P1 is definitely the best, but the bit rate averages about 10 kpbs higher than the others.  It's superiority is clear on this sample http://homepages.apci.net/~yorkgl52/VHCymbals.flac.  This is taken form Van Halen - Atomic Punk.  The whole song is full of flanging like this.  Just for reference, the four encodes of the whole song were:

portable - 158 kpbs
portable1 - 171 kpbs
portable2 - 154 kpbs
portable3 - 159 kpbs

Lame 3.94a13

Reply #18 –
I'd just like to mention that I firmly believe the goal for --preset standard should be virtual TRANSPARENCY, not a certain bitrate. If this version is improved, I would expect it to be a tad lower than the average for 3.90.2. Basically, it shouldn't be ABXable against --ap-s -z in 3.90.2. Please don't throw bits away on making it higher, or save bits in exchange for a certain target.

Lame 3.94a13

Reply #19 –
Quote
I'd just like to mention that I firmly believe the goal for --preset standard should be virtual TRANSPARENCY, not a certain bitrate.

IIRC, MP3 cannot be transparent on some samples, due to codec limitations. Otherwise, I'm sure we'd all agree with you. Admittedly, it is apparently very difficult to ABX -aps and -ape from lossless, just not impossible.

Lame 3.94a13

Reply #20 –
Do you always have this kind of "high" bitrate with your samples? (150-170kbps)

Lame 3.94a13

Reply #21 –
The music I've encoded so far (mostly accoustic instruments, some with lots of percussion) gives 120-130kbps, in a few cases 110-120 or 130-140kbps.
Let's suppose that rain washes out a picnic. Who is feeling negative? The rain? Or YOU? What's causing the negative feeling? The rain or your reaction? - Anthony De Mello

Lame 3.94a13

Reply #22 –
yes. finally vbr for my portable.

but 170 kb seems too much. target should be 130 kbit.

Lame 3.94a13

Reply #23 –
Just out of curiosity, does Alpha 13 include tweaks like -X 1,3 (or 3,3), -Z 1, or shortthreshold for abr presets (eg --preset 160). Thanks.

Lame 3.94a13

Reply #24 –
Van Halen was the highest so far.  System of a Down - Jet Pilot was the lowest.  That had bit rates of 129 with portable,  140 with p1,  123 with p2, and 129 with p3.  P1 was the best of this bunch again, followed by p3, p2, and portable.  Portable sounded really bad, as shown in this clip http://homepages.apci.net/~yorkgl52/JPTest.flac

The classical piece I tried, Stravinsky's Ritual of Abduction didn't have any noticeable artifacts with any of the portables.  The bitrates for this were P - 156, P1 -  155,  P2 -  146,  P3 -  145.

So far, I'd have to say P1 is the best from what I've heard.