IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
Why not and can't add a highpass at 20Hz?
yingmahe
post Sep 15 2008, 03:56
Post #1





Group: Members
Posts: 8
Joined: 1-September 08
Member No.: 57778



Human ears can't notice frenquencies below 20Hz, so why not add a highpass? Won't that save some spaces?
But if i type --highpass 20 it says "highpass frequency too low, highpass disabled".
Please tell me if it is a bug or something else?
Using LAME3.98 -V 6 as i have bad ears. Transparent to me. Don't suprise!
Go to the top of the page
+Quote Post
2E7AH
post Sep 15 2008, 04:24
Post #2





Group: Validating
Posts: 2424
Joined: 21-May 08
Member No.: 53675



--highpass highpass filtering frequency in kHz

This post has been edited by 2E7AH: Sep 15 2008, 04:25
Go to the top of the page
+Quote Post
Rio
post Sep 15 2008, 05:19
Post #3





Group: Members
Posts: 207
Joined: 18-December 06
From: Olongapo
Member No.: 38799



I understand the dilemma of the OP. If one would want to disable frequencies below 20Hz, and the highpass filter should be in kHz, then it would be a problem.

Expect to hear that -Vx switches are there for a reason, and hence, no need for additional switches.


--------------------
"Listen to me...
Never take unsolicited advice..."
Go to the top of the page
+Quote Post
Canar
post Sep 15 2008, 05:49
Post #4





Group: Super Moderator
Posts: 3373
Joined: 26-July 02
From: To:
Member No.: 2796



Frequencies below 20Hz require very, very few bits to encode. I'd be surprised if such a highpass filter cut 1kbps off the overall size of the MP3.

Edit: Removed typoed "k" in front of "Hz"

This post has been edited by Canar: Nov 13 2008, 16:55


--------------------
You cannot ABX the rustling of jimmies.
No mouse? No problem.
Go to the top of the page
+Quote Post
yingmahe
post Sep 15 2008, 06:02
Post #5





Group: Members
Posts: 8
Joined: 1-September 08
Member No.: 57778



Thanks for replies guys.
I just tried --highpass 0.02 it didn't work either.
Anyway I'm happy with -V 6.

And I have another question:
I've recently bought some MP3 from Amazon and they're 256kbps. For portable use, I want to cut the file size down, I'm thinking transcode them to -V 6, will that harm the quality a lot? Or I should transcode to -V 5 to get the same quality from WAV to -V 6?
Go to the top of the page
+Quote Post
Canar
post Sep 15 2008, 06:06
Post #6





Group: Super Moderator
Posts: 3373
Joined: 26-July 02
From: To:
Member No.: 2796



Only you're qualified to make that assessment, yingmahe. Give it a listen, see how it sounds. High-bitrate to low-bitrate conversions don't hurt that much.


--------------------
You cannot ABX the rustling of jimmies.
No mouse? No problem.
Go to the top of the page
+Quote Post
mikenet
post Sep 15 2008, 06:08
Post #7





Group: Members
Posts: 39
Joined: 15-June 02
Member No.: 2313



Exactly. There's only 20Hz of bandwidth between 0Hz and 20Hz, or about 0.09% of the bandwidth of CD Audio. That portion of the spectrum only takes up 1.28kbps in CD Audio, before any compression takes place. Probably not worth worrying about in an MP3.

EDIT: Whoops, too slow wink.gif.

This post has been edited by mikenet: Sep 15 2008, 06:09
Go to the top of the page
+Quote Post
Nick.C
post Sep 15 2008, 07:21
Post #8


lossyWAV Developer


Group: Developer
Posts: 1813
Joined: 11-April 07
From: Wherever here is
Member No.: 42400



Surely a 20Hz sine wave at 44.1kHz sample rate will still require 1411.2kbit/s?


--------------------
lossyWAV -q X -a 4 --feedback 4| FLAC -8 ~= 320kbps
Go to the top of the page
+Quote Post
knutinh
post Sep 15 2008, 07:50
Post #9





Group: Members
Posts: 570
Joined: 1-November 06
Member No.: 37047



QUOTE (Nick.C @ Sep 15 2008, 08:21) *
Surely a 20Hz sine wave at 44.1kHz sample rate will still require 1411.2kbit/s?

Not if it is compressed using WinRAR, WinZIP or something similar.

http://en.wikipedia.org/wiki/Information_theory

A 20Hz sinewave could moreover be resampled to a samplingrate of >40Hz, meaning that the bitrate would be divided by 1000.

-k

This post has been edited by knutinh: Sep 15 2008, 07:52
Go to the top of the page
+Quote Post
yingmahe
post Sep 15 2008, 11:45
Post #10





Group: Members
Posts: 8
Joined: 1-September 08
Member No.: 57778



OK thanks everybody, planning to do some tests soon.
Go to the top of the page
+Quote Post
pdq
post Sep 15 2008, 12:58
Post #11





Group: Members
Posts: 3450
Joined: 1-September 05
From: SE Pennsylvania
Member No.: 24233



There was a recent thread that discussed the lowest settable value for the highpass filter. I think the conclusion was something on the order of 400Hz.

Edit: found the thread here. smile.gif

This post has been edited by pdq: Sep 15 2008, 17:47
Go to the top of the page
+Quote Post
yingmahe
post Sep 15 2008, 13:32
Post #12





Group: Members
Posts: 8
Joined: 1-September 08
Member No.: 57778



Who will want a highpass at 400Hz? That'll filter most bass!
Go to the top of the page
+Quote Post
coincoin
post Oct 5 2008, 01:35
Post #13





Group: Members
Posts: 1
Joined: 5-October 08
Member No.: 59361



A highpass at 20 Hz would be a bad idea. Most modern records are sent through a limiter, and as such removing the low end would actually increase the peak level! So, you will need to reduce the level to avoid clipping, and you might even have to spend more bits to encode the low end due to higher required PSNR...

Not to mention that you wouldn't gain anything because commercial releases don't have insane levels of sub-20 Hz content, and because the MP3 frequency resolution is much too coarse to be able to "notice" that there is nothing below 20 Hz.
Go to the top of the page
+Quote Post
pdq
post Oct 5 2008, 13:37
Post #14





Group: Members
Posts: 3450
Joined: 1-September 05
From: SE Pennsylvania
Member No.: 24233



QUOTE (coincoin @ Oct 4 2008, 20:35) *
A highpass at 20 Hz would be a bad idea. Most modern records are sent through a limiter, and as such removing the low end would actually increase the peak level! So, you will need to reduce the level to avoid clipping, and you might even have to spend more bits to encode the low end due to higher required PSNR...

Huh!? blink.gif

The limiter was applied before recording the CD, so any reduction of the higher frequencies has already occurred. Later removal of low frequencies has no effect on the amplitude of the higher frequencies.

As for encoding of frequencies near or below 20 Hz, only a very small fraction of the total bits are used for encoding such low frequencies
Go to the top of the page
+Quote Post
2Bdecided
post Oct 6 2008, 16:15
Post #15


ReplayGain developer


Group: Developer
Posts: 5360
Joined: 5-November 01
From: Yorkshire, UK
Member No.: 409



It's even less beneficial to high pass filter than this discussion suggests.

All frequencies below 20Hz (and quite a few above that - double that even) are going to land in the DC bin - i.e. you don't even fit a full cycle within one mp3 frame.

I'm guessing that most mp3 encoders will be reasonably careful with the DC bin - not because we can hear DC, but because systematic or random changes in DC level from frame-to-frame would cause horrible discontinuities if the blocks weren't overlapped. Since they are overlapped, what you would get is a sinusoidal-like waveform at frame rate. That's 44.1k/576=77Hz for long blocks or 44.1k/192=230Hz for short blocks. Not nice.


You'd need to high pass filter at a much higher frequency (e.g. above the two calculated above!) to ensure the DC bin was always approximately zero. So they'll usually be content in that bin, and it'll need to be encoded quite accurately - removing a little of the content below 20Hz will make no useful difference.

(Having said that, a large DC offset is not a good thing on an audio signal, but if present, it should be removed with care).

Cheers,
David.
Go to the top of the page
+Quote Post
Paul Sanders
post Oct 12 2008, 20:55
Post #16





Group: Members
Posts: 104
Joined: 19-May 08
From: UK (London-ish)
Member No.: 53626



For what it's worth, I have made a number of vinyl recordings using VBR, quality 0 (= maximum), and rumble of around 30Hz is clearly visible on the lead-in of most of them. Ergo, MP3 files can carry information at these very low frequencies, at high bit rates certainly.

Unless, and I'm really not sure about this, these are artefacts. Would anybody know if this might be the case? I have not re-recorded any of these albums to a lossless format, unfortunately.

This post has been edited by Paul Sanders: Oct 12 2008, 20:55


--------------------
I am an independent software developer (VinylStudio) based in UK
Go to the top of the page
+Quote Post
audiobookie
post Nov 13 2008, 07:05
Post #17





Group: Members
Posts: 22
Joined: 16-January 07
Member No.: 39699



I was trying to get highpass into a command line earlier this year.

http://www.hydrogenaudio.org/forums/index....c=64185&hl=

I for one would love Highpass filters to work below 400Hz.
Very useful for encoding audiobooks were there is just the human voice envolved.
Even if the saving was small, every little byte helps over hours of audio.....
Go to the top of the page
+Quote Post
SebastianG
post Nov 13 2008, 18:07
Post #18





Group: Developer
Posts: 1318
Joined: 20-March 04
From: Göttingen (DE)
Member No.: 12875



QUOTE (2Bdecided @ Oct 6 2008, 17:15) *
It's even less beneficial to high pass filter than this discussion suggests.

All frequencies below 20Hz (and quite a few above that - double that even) are going to land in the DC bin - i.e. you don't even fit a full cycle within one mp3 frame.

Due to the nature of the filterbank (polyphase filterbank + MDCT with sine window) a DC offset will affect only the first MDCT coefficient. Other low frequencies affect a group of transform coefficients each due to the spectral leakage effect.

To avoid excessively large low frequency coefficients the A/52 (Dolby Digital) specification even suggests the use of a DC-removing 2nd order Butterworth high pass filter -- if I recall correctly. I don't think it's a bad idea to apply this kind of filter prior MP3 encoding. On "normal" audio (with no DC offset / rumble) this should not make a big difference w.r.t. quality/bitrate.

Cheers!
SG
Go to the top of the page
+Quote Post
audiobookie
post Nov 14 2008, 02:11
Post #19





Group: Members
Posts: 22
Joined: 16-January 07
Member No.: 39699



Looking back again at my old thread I see it was 350 not 400 Hz.
Still a useless frequency setting no one would want to use.....

Looking at the voice recording I have, nothing is really going on under 100 Hz.
But maybe this is ignored by Lame anyway and doesn't use many bits.
Go to the top of the page
+Quote Post
Gabriel
post Dec 2 2008, 13:53
Post #20


LAME developer


Group: Developer
Posts: 2950
Joined: 1-October 01
From: Nanterre, France
Member No.: 138



Due to the way Lame's filtering is implemented, it can not offer such granularity, which is why very low values get ignored.
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: 20th December 2014 - 02:39