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: What about vbr-old -V1 in Lame 3.91 ? (Read 3961 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

What about vbr-old -V1 in Lame 3.91 ?

In Lame 3.89 I used simple settings:
-V1 -mj -q0 -b112  (--lowpass 19.5 or with -k )

But it seems that in 3.91 vbr-old is worse than in 3.89 - I tested on castanets.wav and fatboy.wav.  Sorry - now I have not more time for full testing...  And I have some questions against Lame 3.91:

1) Why does Lame 3.91 use ATH 4 instead ATH 2 in vbr-old ?

2) Why does --alt-preset standard  use --lowpass 18.7 instead 19.5 ?

3) May I use --lowpass 20 or -k  with --alt-preset standard ?

And general:
4) What about  -V1 -mj -q0 -b112  vs  --alt-preset standard  (that use -V2 -q2) ???


Thank you!

What about vbr-old -V1 in Lame 3.91 ?

Reply #1 –
Quote
Originally posted by toader
But it seems that in 3.91 vbr-old is worse than in 3.89 - I tested on castanets.wav and fatboy.wav.  Sorry - now I have not more time for full testing...  And I have some questions against Lame 3.91:[


If you find some time for full testing it might be interesting to see your results.  Otherwise I'd have to disagree with you.  The quality of 3.91 is significantly higher than that of 3.89.

Quote
1) Why does Lame 3.91 use ATH 4 instead ATH 2 in vbr-old ?


Because the LAME developers decided this was the approach they wanted to use.  ATH type 4 is basically a more efficient and elegant approach to gradually degrade quality and bitrate as a higher V setting is used.

Quote
2) Why does --alt-preset standard  use --lowpass 18.7 instead 19.5 ?


Because it saves bits and because so far nobody has been able to hear the difference to my knowledge.  If someone can reliably hear the difference across a variety of samples and they can prove this via ABX, I will obviously reconsider this decision, if not, then it's a non-issue.

Quote
3) May I use --lowpass 20 or -k  with --alt-preset standard ?


Yes, but unless you can reliably hear the difference (this is probably doubtful) and can verify via abx, then you'd be wasting bits, especially when using -k.  For that matter, using -k could result in a reduction in quality so I wouldn't recommend it.

Quote
And general:
4) What about  -V1 -mj -q0 -b112  vs  --alt-preset standard  (that use -V2 -q2) ???


All the V settings do is globally adjust masking in one direction or the other.  I have found that this is not the best approach to increase quality and that V2 is the "sweet spot" upon which to build --alt-preset standard.  The techniques --alt-preset standard uses have a significantly greater effect than using a higher V setting which would basically just end up wasting bits.

-q0 is not used because it enables Takehiro's new noise shaping method which allows for more room for error (possibly lower quality) and a 40-400% reduction in speed.  All you gain is about 2-5kbps lower bitrate overall.  So basically you now have the possibility for:

1.  Lower quality.
2.  A massive speed hit.
3.  About 2-5kbps lower bitrate.

I'd say that should be reason enough not to use it.

What about vbr-old -V1 in Lame 3.91 ?

Reply #2 –
Quote
Originally posted by toader
And general:
4) What about  -V1 -mj -q0 -b112  vs  --alt-preset standard  (that use -V2 -q2) ???

This has been explained many times... alt-preset standard/extreme/insane use code level tweaks not accessible by external switches. So external switches comparing non-alt-preset and alt-preset like you show above are not comparable by any means. And sure you can use different lowpass with --alt-presets if you want, but I would check if there's any use with abx tests. It's certainly not easy to abx reliably between 18.7 and 19.5, if not impossible.

Second.. -q0 uses now experimental noise shaping (which was not in 3.89) and I would be very causious. It uses more cutting edge noise shaping which means more possibilities to fail. I would use -q2. This might be one reason why you said 3.90 vbr-old is worse..
Read this thread: http://www.hydrogenaudio.org/forums/showth...s=&threadid=378
Juha Laaksonheimo

What about vbr-old -V1 in Lame 3.91 ?

Reply #3 –
Nice Dib.. Should have guessed you are faster again and my message was in vain...
Juha Laaksonheimo

What about vbr-old -V1 in Lame 3.91 ?

Reply #4 –
Thank you for answer !!! 

I am trying blind audio tests already.
What program may I use for blind random rename mp3 files ?  Because it is very uneasy to blind rename by hands :-)
 

What about vbr-old -V1 in Lame 3.91 ?

Reply #5 –
Quote
Originally posted by JohnV
Nice Dib.. Should have guessed you are faster again and my message was in vain...


Lol! Sorry

Quote
Originally posted by toader
What program may I use for blind random rename mp3 files ? Because it is very uneasy to blind rename by hands :-)


You don't really need to rename the files, you can just use a program which randomizes the files on playback.  Check out the program at www.pcabx.com, this is what is considered as the "standard" around here.  Soon sphoid and I will be releasing our cross platform utility which will be similar to that tool but much more advanced..