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 3.94a15 (Read 13150 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Lame 3.94a15

New from a14:

*quality fix (a bug already corrected during a14)
*lame -b xxx / --abr xxx are using presets
*some fixes in the INFO tag
*speedup


This one should be close to the release (at least I hope).
What will still change are the low vbr presets ( lower than 150kbps), which were non-existing before.

Lame 3.94a15

Reply #1 –
Where can I download the Lame3.94a15? I tried from LAME Binaries site but the latest is the 14 alpha.

Lame 3.94a15

Reply #2 –
tomorrow from the usual websites

Lame 3.94a15

Reply #3 –
Or sooner if John33 sees this.
Juha Laaksonheimo

Lame 3.94a15

Reply #4 –
Someone should probably try to coordinate the testing.
IMHO the interesting points would be:
1. Quality of --preset (fast) medium/standard/extreme
compared to 3.90.3(modified)
2. Quality of 128kbps default cmd.line
compared to 3.90.3 --alt-preset (cbr) 128

Bring your own samples or use known problem ones for the higher bitrates and let's finally move on from 3.90.3.

dev0
"To understand me, you'll have to swallow a world." Or maybe your words.

Lame 3.94a15

Reply #5 –
CVS is still showing 3.94.14 alpha in version.h. Is the update committed, or does that need an update?

Lame 3.94a15

Reply #6 –
Anonymous cvs from sourceforge has a one day delay.

Lame 3.94a15

Reply #7 –
Quote
Anonymous cvs from sourceforge has a one day delay.

Sometimes, it has almost a week delay.

Lame 3.94a15

Reply #8 –
Quote
Anonymous cvs from sourceforge has a one day delay.

You'll have to wait until tomorrow then, folks!!

Lame 3.94a15

Reply #9 –
Quote
You'll have to wait until tomorrow then, folks!!

Else, Gabriel could make you Lame developer

Lame 3.94a15

Reply #10 –
so glad to see that the development makes progress! I can't wait for the final release, that will hopefully stop all the confusion about the different releases of LAME! 
--alt-presets are there for a reason! These other switches DO NOT work better than it, trust me on this.
LAME + Joint Stereo doesn't destroy 'Stereo'

Lame 3.94a15

Reply #11 –
are there any regression tests  done with current alphas?

(is the rumour true that JohnV is doing that behind the scenes?)

are we supposed to do any?

EDIT: grammar

Lame 3.94a15

Reply #12 –
Quote
I can't wait for the final release, that will hopefully stop all the confusion about the different releases of LAME!

No, it's just one more 
In theory, there is no difference between theory and practice. In practice there is.

Lame 3.94a15

Reply #13 –
Is the issue with INFO tags and inaccurate padding value fixed ? LAME 3.90.* had a problem with padding values being wrong with certain frontends (probably encoding from stdin / WAV headers created "on the fly" not having correct length info in their headers, confusing LAME), didn't happen with plain lame.exe commandline encoding file.wav => file.mp3.
Microsoft Windows: We can't script here, this is bat country.

Lame 3.94a15

Reply #14 –
Quote
are there any regression tests  done with current alphas?

(is the rumour true that JohnV is doing that behind the scenes?)

Wtf rumors, I haven't heard anything.. 

Truth is I haven't touched Lame alphas for months, and it seems unlikely that I have much time in the future for testing it. Maybe if the development was faster and more goal oriented..

But I hope that some other people are eager to give feedback to the developers.
Juha Laaksonheimo

Lame 3.94a15

Reply #15 –
YES!  I came here to bitch about 3.94 and here i see Gabriel is ahead of me.

Post a binary and I'll try and help.

Lame 3.94a15

Reply #16 –
Quote
Is the issue with INFO tags and inaccurate padding value fixed ? LAME 3.90.* had a problem with padding values being wrong with certain frontends (probably encoding from stdin / WAV headers created "on the fly" not having correct length info in their headers, confusing LAME), didn't happen with plain lame.exe commandline encoding file.wav => file.mp3.


I have to check this. When using stdin (pipes) Lame is not able to write into the Info tag after the encoding is finished. It means that it is not possible to have the padding value, but at least it should be 0.

Lame 3.94a15

Reply #17 –
Binaries now available at Rarewares.

Lame 3.94a15

Reply #18 –
After reading the history, I have some questions.

* best huffman divide in the inner loop. This should improve the quality, but PAINFULLY slow. So it is not enabled by default. Use -q0 to use it.

Does this mean, that 3.94 allows the combination of -aps and -q0, to improve the quality?

* Changed -q option mapping. "-q2" until version 3.93 is now "-q3".
Is q1 now q2 and q0 now q1? Or: what's the new q2?

Lame 3.94a15

Reply #19 –
Quote
Does this mean, that 3.94 allows the combination of -aps and -q0, to improve the quality?


3.94 allows the combination of aps an q0, but this will not improve quality. It will only reduce file size. You have to use abr/cbr in order for it to improve quality.

Quote
* Changed -q option mapping. "-q2" until version 3.93 is now "-q3".
Is q1 now q2 and q0 now q1? Or: what's the new q2?


q3 is the default.
q2 is adding substep_shaping 2
q1 is adding noise_shaping_amp 2
q0 is adding best_huffman 2


Btw you can use --verbose to know all this.

Lame 3.94a15

Reply #20 –
Quote
3.94 allows the combination of aps an q0, but this will not improve quality. It will only reduce file size. You have to use abr/cbr in order for it to improve quality.

Out of curiosity I just tested lame 3.94a15 --alt-preset standard vs. --alt-preset standard -q 0 with one file. Result:
Encoding speed: 3.8x vs. 0.8x
Size: 1.284KB vs. 1.291KB (-q 0 bigger  )

Is this bad luck or ...?

Wave substraction -aps - -aps-q0 : CEP statistics:
Code: [Select]
    Left    Right
Min Sample Value:    -136    -167
Max Sample Value:    105    157
Peak Amplitude:    -47.64 dB    -45.85 dB
Possibly Clipped:    0    0
DC Offset:    0  0 
Minimum RMS Power:    -105.18 dB    -106.44 dB
Maximum RMS Power:    -60.46 dB    -59.51 dB
Average RMS Power:    -79.17 dB    -78.6 dB
Total RMS Power:    -76.41 dB    -75.66 dB
Actual Bit Depth:    16 Bits    16 Bits

Using RMS Window of 50 ms

This means the differences between -aps and -aps -q0 are more than just size. So listening tests would be needed to judge if there are quality differences IMO. Correct me if I'm wrong please.
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 3.94a15

Reply #21 –
Quote
Size: 1.284KB vs. 1.291KB (-q 0 bigger  )

Is this bad luck or ...?

I think this is a bad luck case

Lame 3.94a15

Reply #22 –
.... is this a frozen topic???

Lame 3.94a15

Reply #23 –
Quote
.... is this a frozen topic???

The last message (before yours) was posted today...
what are you talking about ? 

Lame 3.94a15

Reply #24 –
Is this a testing thread ??
Nevertheless...i’ve done some quick comparison with --alt-preset cbr 128 3.90.3 vs 3.94a15 using applaud.wav and campestre.wav (already used here). Both samples are worsened with the new alpha, there is a sort of ringing-watery sound.
As already pointed out by JohnV (when confirming the old 394a12 bug) a spectral analisys show some (sometimes small) dropouts if compared with 3.90.x releases. These problems, that are audible in at least these two samples, are very evident with campestre.wav. Look at this picture swowing 3.90.3 (at top) vs. 3.94a15.

The adoption of some alt-presets by default, even if tardive and with a slightly modified behaviour, is really a good news. As dev0 already said, is time to organize the testing process. In my spare time i can give my very humble contribute and i think that here at HA there are many other people that can contribute as well. I think that Gabriel or someone else LAME developer should address the testing in some way telling us what settings should be tested with the specified alpha, in this manner the testing is much more easy and addressed by someone much more competent.
What do you think about ?
WavPack 4.3 -mfx5
LAME 3.97 -V5 --vbr-new --athaa-sensitivity 1