IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
Adaptive Lowpass in Rev6 of lame_dm??
RD
post Nov 17 2001, 22:13
Post #1





Group: Members
Posts: 151
Joined: 29-September 01
Member No.: 31



I was doing some testing with lame_dm rev6 and I noticed on one of my files, a pure percussion .wav with lots of highhats and cymbals, that the lowpass numbers, usually 18671- 19205 INCREASED to 18774-19355.

I did not alter the --alt-preset normal commandline so my first question is:

1. Did Dibrom implement an adaptive lowpass?

2. If yes can one turn it on using switches? or does it only run internally?

3. If no, how did this happen?

4. Is there any other way to increase the lowpass to 18774 with other versions of lame or no? --lowpass width maybe?

Thanks.
Go to the top of the page
+Quote Post
ancl
post Nov 17 2001, 22:31
Post #2





Group: Members (Donating)
Posts: 185
Joined: 29-September 01
Member No.: 54



My guess is that the different sample rate (48 vs 44.1) may have something to do with this. biggrin.gif

/Andreas
Go to the top of the page
+Quote Post
RD
post Nov 17 2001, 22:37
Post #3





Group: Members
Posts: 151
Joined: 29-September 01
Member No.: 31



:jawdrop: Doh!!!!!!!! I'm such a fool! :insane:

You are right!

Well now that I have embarrassed my self I might as well ask two more questions...

1. Is implementing an adaptive lowpass possible? and advanatageous?

2. Why are certain values fixed? For example who decided on 18671 for lowpass 19? Can we edit the source code to make it 19000 for lowpass 19?

Anybody???
Go to the top of the page
+Quote Post
Dibrom
post Nov 17 2001, 23:45
Post #4


Founder


Group: Admin
Posts: 2958
Joined: 26-August 02
From: Nottingham, UK
Member No.: 1



QUOTE
Originally posted by RD
1.  Is implementing an adaptive lowpass possible? and advanatageous?


Yes, and Yes. Can I do it? Not currently. But I think an adaptive lowpass could solve quite a few problems... I think this is one of the next *big* things that LAME needs..

QUOTE
2.  Why are certain values fixed? For example who decided on 18671 for lowpass 19?  Can we edit the source code to make it 19000 for lowpass 19?


Well part of the reason is because its not a hard lowpass.. the polyphase part for example. It will cut off nearly completely around 19khz though.
Go to the top of the page
+Quote Post
RD
post Nov 18 2001, 02:30
Post #5





Group: Members
Posts: 151
Joined: 29-September 01
Member No.: 31



I know there are things concerning ease of use, frontends, etc. that you want to change....BUT What other big things do you think Lame needs concerning QUALITY?
Go to the top of the page
+Quote Post
Dibrom
post Nov 18 2001, 02:40
Post #6


Founder


Group: Admin
Posts: 2958
Joined: 26-August 02
From: Nottingham, UK
Member No.: 1



QUOTE
Originally posted by RD
BUT What other big things do you think Lame needs concerning QUALITY?


Well the adaptive lowpass for one, and a much better noise measuring system (better tonality estimation helps here) and method for determining best quantization... (this last part is what I'm messing around with myself at the moment..). Some sort of better impulse detection would be nice also.. and a 2-pass mode for abr.
Go to the top of the page
+Quote Post
JohnV
post Nov 19 2001, 18:06
Post #7





Group: Developer
Posts: 2797
Joined: 22-September 01
Member No.: 6



Naoki is also testing new noise shaping for Lame. One thing I wonder is that is the Lame's spreading function really optimal. It's said that it has been done using listening tests, but I wonder who were the testers.

Anyway if Naoki can make new noise shaping and the more difficult part - new tonality estimations to work, it could mean that achieving good quality/bitrate (profile tweking) becomes much more easier. Anyway, with optimal spreading function and tonality estimations, nearly optimal masking threshold and overall better simultaneous masking could be achieved.
Dibrom's current tweking is working pretty well, but even he thinks it's pretty roughly done, meaning that several special situations must be tweaked via code level forking, and this does lead to a bit non-optimal bitrates sometimes especially with some low volume music.


--------------------
Juha Laaksonheimo
Go to the top of the page
+Quote Post
RD
post Nov 20 2001, 15:15
Post #8





Group: Members
Posts: 151
Joined: 29-September 01
Member No.: 31



JohnV,

Any idea of a time frame on when Naioki might be finished with the new tonality estimation, etc?

I mean are we talking many months or just weeks?
Go to the top of the page
+Quote Post

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: 23rd August 2014 - 08:52