IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
Replay Gain 89 db reference, What does it stand for ?
Joop
post Aug 14 2003, 22:28
Post #1





Group: Members
Posts: 4
Joined: 14-August 03
Member No.: 8366



I have read through the forum but still have my question..

What is the 89 dB "no clipping" level actually stand for? In audio devices 0 dB is equal to 0,775 or 1 V, but what is it in the informatica world?

and very much related to this, i don't understand why 89 should not clip and 100 db should. I've read that clipping would occur at decoding to wav.. but i don't do that.
So then it would be ok to encode to 100 dB? or does the digital mp3 data clip at this level? or the audio cards output channels? and what would be the output voltage level of the soundcard at 89 or 100 dB, if 0 dB is already 0,775 V? Greetings and thanks if you can provide a solid answer..

This post has been edited by Joop: Aug 14 2003, 22:29
Go to the top of the page
+Quote Post
AstralStorm
post Aug 14 2003, 23:33
Post #2





Group: Members
Posts: 745
Joined: 22-April 03
From: /dev/null
Member No.: 6130



It is dB of dynamic range (steps), not the one of voltage or pressure - you're certainly mixing these up.
Soundcards accept line level voltages without clipping - how they map it to digital values depends on ADC and volume setting.
Digital data has always limited dynamic range - loudest possible sound is mapped at 0 dB,
most quiet usually -96 dB (16 bit) or -144 dB (24 bit).
Any sound exceding 0 dB (digital) will clip and it doesn't sound nice most of the times,
while any quieter qound will be truncated (dithering can exploit some properties of our hearing
and allow us to hear even quieter sound even if it's not there)
Of couse, how loud is that sound depends only on amplifier settings,
as these values are only dynamic range, not loudness.


--------------------
ruxvilti'a
Go to the top of the page
+Quote Post
Joop
post Aug 15 2003, 22:26
Post #3





Group: Members
Posts: 4
Joined: 14-August 03
Member No.: 8366



Thank you very much for your reply. But I missed the clue about the 89 dB?

Now I have installed a program which shows an analog VU meter on my computer schreen (with Winamp, from Boris Ribov). It has just like the real VU meter a reading until 0 dB, and above that the schale gets red to indicate distortion.
A test mp3 file goes with it and it shows a level of 0 dB.
Now mp3gain shows this mp3 file at a level of 100 dB.
So is it just a matter of offset, that is when mp3gain talks about 89 dB it should be -11 dB? What is the reason for having two expressions for the same thing?

Also there seems to be a "global gain field" in mp3, 8 bits, controlling the loudness of each individual frame. According to mp3gain is the db level related to this frame as 1.5 dB per step. With 255 steps this gives a global dynamic range of 384 dB!

Which such a range, i would think the "0 dB" would be selected somewhere in the middle and not at the maximum sound level. This would explain why Boris VU meter happily exceeds the 0 dB up to +8 dB, and why mp3gain can claim the change is lossless (the chance the global gain range limit is reached is very small).
But, i still dont understand from this the statement "clipping can occur when 89 dB is exceeded".

Perhaps nice to know, of the 2000 songs I have processed with mp3gain, 1100 were not changed, 130 were increased, and 770 were decreased (with max 12 dB).

Maybe you can shine some light on this (not from New York I hope?)
Go to the top of the page
+Quote Post
Snelg
post Aug 16 2003, 00:50
Post #4


MP3Gain Developer


Group: Developer
Posts: 207
Joined: 30-December 01
Member No.: 846



QUOTE (Joop @ Aug 15 2003, 02:26 PM)
Thank you very much for your reply. But I missed the clue about the 89 dB?

The confusion here is about what "dB" means. dB is just a logarithmic scale and can refer to just about anything you want it to.
The dB in VU meters and such refers to the amplitude. 0dB is the highest non-clipping amplitude.

The dB in MP3Gain is referring to the Sound Pressure Level (SPL), not amplitude.

Two different files with the same peak amplitude (i.e. both hit the same spot on a VU meter) can have very different "volumes" (i.e. how loud they sound to your ears).

Read David Robinson's original Replay Gain proposal for more in-depth info.
http://replaygain.hydrogenaudio.org

This page specifically explains the 89dB

http://replaygain.hydrogenaudio.org/calibration.html

(actually, that page uses David's original target of 83dB, which was deemed too quiet by users of early versions of mp3gain)

-Glen
Go to the top of the page
+Quote Post
Roosh
post Aug 16 2003, 04:46
Post #5





Group: Members
Posts: 32
Joined: 12-August 03
Member No.: 8334



Apologies as I may be throwing off the rhythm of this thread (thought these questions didn't deserve to be in a new thread) and am definitely inserting some very noobish qualities to it, but unsure.gif :

1)Is it possible to adjust ape or mpc files w/ Mp3 Gain (or do I even need to)?

2)I currently have version 1.0.0.0 (or 1.3.3 -- which is it?) The beginning of this thread from a couple of years ago makes reference to "radio" settings/radio analysis which I don't find in my version. Is this what I want for an update? "mp3gain-1_4_2-src"

Ok, thanks.
Go to the top of the page
+Quote Post
Snelg
post Aug 16 2003, 05:53
Post #6


MP3Gain Developer


Group: Developer
Posts: 207
Joined: 30-December 01
Member No.: 846



QUOTE (Roosh @ Aug 15 2003, 08:46 PM)
2)I currently have version 1.0.0.0 (or 1.3.3 -- which is it?) The beginning of this thread from a couple of years ago makes reference to "radio" settings/radio analysis which I don't find in my version. Is this what I want for an update? "mp3gain-1_4_2-src"

Yeah, I think this thread should be locked or something. Every few weeks/months someone comes along and asks another random MP3Gain question in it...

Anyhow, the word "Radio" was replaced by "Track" at some point, because it made more sense in context.

I'm assuming you're using the GUI (lovely Windows program, not type-it-all-yourself DOS). You want "mp3gain-win-1_2_1" (that's GUI version 1.2.1, which also contains DOS version 1.4.2-- not to be confusing or anything...)

-Glen
Go to the top of the page
+Quote Post
Roosh
post Aug 16 2003, 07:24
Post #7





Group: Members
Posts: 32
Joined: 12-August 03
Member No.: 8334



Sorry (your section at sourceforge isn't really self-explanatory for a newbie) & thanks. wink.gif
Go to the top of the page
+Quote Post
Joseph
post Aug 16 2003, 07:47
Post #8





Group: Members (Donating)
Posts: 108
Joined: 7-April 03
From: Newark, CA
Member No.: 5871



Dang, I wish I knew about this stuff back when I first got my Rio 500. smile.gif
Go to the top of the page
+Quote Post
Snelg
post Aug 16 2003, 08:09
Post #9


MP3Gain Developer


Group: Developer
Posts: 207
Joined: 30-December 01
Member No.: 846



QUOTE (Roosh @ Aug 15 2003, 11:24 PM)
Sorry (your section at sourceforge isn't really self-explanatory for a newbie) & thanks. wink.gif

Check out the just-uploaded
http://mp3gain.sourceforge.net/download.php

That should be much clearer.

-Glen
Go to the top of the page
+Quote Post
Pio2001
post Aug 16 2003, 09:55
Post #10


Moderator


Group: Super Moderator
Posts: 3936
Joined: 29-September 01
Member No.: 73



89 db is also called K-14. It means that the average perceived level is set at -14 db.
89 db refers to professional calibrated studio monitors : when a -14 db digital file is played (the reference is a pink noise), if the ampli gain is properly calibrated, the sound pressure must be 89 db at the listening point.
Above this, you risk clipping, because if somewhere in the track there is a peak that is more than 14 db above the average level of the track, then it will be pushed above 0db and will clip.
Go to the top of the page
+Quote Post
Joop
post Aug 17 2003, 13:13
Post #11





Group: Members
Posts: 4
Joined: 14-August 03
Member No.: 8366



still have a lot to learn..

Hydrogen explains the replaygain data format, 9 bits plus sign, with max range + and - 51,2 dB (512/10); but i understand this format is outdated by the new direct mp3 adjustment (256 * 1.5 dB)?

but maybe this is the clue i'm looking for: the conversion from mp3 to wav happens every time you play the file (i thought i wouldn't do decoding) .
Then an mp3 file mp3gained to 100 dB would still have no clipping but the moment it is played it causes clipping at the digital-to-analog decoder.

But then you could also adjust the replay volume to -11 dB for maximum non-clipping output. For this to work the Winamp software should first adjust the volume and then decode the mp3 to wav and on to analog.

Why i go through all this?

Well, from the analog recorder days (still have one of those fantastic Revox A77 tape recorders) i'm used to accept some red area. I hear no problems with it.
The recorder does not have such a signal-to-noise ratio to allow the comfort of completely no clipping..
Digital recording seem to copy this strategy as it has a red area as well. So to me the sound level seems ok if it sometimes hits the red area.
And, i'm used to limit the play-back volume to max 70%. The remainder seems reserved to compensate for accidentely faint recordings.

Please tell me if i'm on the right track.

This post has been edited by Joop: Aug 17 2003, 13:18
Go to the top of the page
+Quote Post
Roosh
post Aug 17 2003, 19:35
Post #12





Group: Members
Posts: 32
Joined: 12-August 03
Member No.: 8334



QUOTE
1)Is it possible to adjust ape or mpc files w/ Mp3 Gain (or do I even need to)?
Go to the top of the page
+Quote Post
CiTay
post Aug 17 2003, 19:45
Post #13


Administrator


Group: Admin
Posts: 2378
Joined: 22-September 01
Member No.: 3



QUOTE (Roosh @ Aug 17 2003, 08:35 PM)
QUOTE
1)Is it possible to adjust ape or mpc files w/ Mp3 Gain (or do I even need to)?

MP3Gain is for MP3 only. Replaygain for APE files: Only indirectly via WaveGain. But that's not lossless. For MPC: Yes, with ReplayGain for MPC (see recommended MPC settings and that forum section in general).
Go to the top of the page
+Quote Post
AstralStorm
post Aug 17 2003, 20:58
Post #14





Group: Members
Posts: 745
Joined: 22-April 03
From: /dev/null
Member No.: 6130



You've forgot Foobar's ability to do replaygain on anything.
(APE files too - it writes info into APEv2 tags)

This post has been edited by AstralStorm: Aug 17 2003, 21:00


--------------------
ruxvilti'a
Go to the top of the page
+Quote Post
CiTay
post Aug 17 2003, 21:13
Post #15


Administrator


Group: Admin
Posts: 2378
Joined: 22-September 01
Member No.: 3



QUOTE (AstralStorm @ Aug 17 2003, 09:58 PM)
You've forgot Foobar's ability to do replaygain on anything.

Hm yeah... MA should come bundled with foobar laugh.gif
Go to the top of the page
+Quote Post
Volcano
post Aug 17 2003, 21:48
Post #16





Group: Members (Donating)
Posts: 916
Joined: 30-September 01
From: Berlin, Germany
Member No.: 112



QUOTE (AstralStorm @ Aug 17 2003, 09:58 PM)
You've forgot Foobar's ability to do replaygain on anything.
(APE files too - it writes info into APEv2 tags)

Unfortunately, foobar2000 is the only player that supports this kind of thing for Monkey's Audio files as of yet - and from what I read somewhere here, it doesn't seem likely that the Monkey's Audio Winamp plugin will feature RG support. IIRC, it was said that Matt considers ReplayGain the job of an audio player, not of the format itself - which is plain idiotic if you ask me, but then I'm no big fan of Monkey's Audio anyway. dry.gif

This post has been edited by Volcano: Aug 17 2003, 21:48
Go to the top of the page
+Quote Post
2Bdecided
post Aug 18 2003, 13:50
Post #17


ReplayGain developer


Group: Developer
Posts: 5364
Joined: 5-November 01
From: Yorkshire, UK
Member No.: 409



QUOTE (Joop @ Aug 17 2003, 12:13 PM)
Well, from the analog recorder days (still have one of those fantastic Revox A77 tape recorders) i'm used to accept some red area. I hear no problems with it.
The recorder does not have such a signal-to-noise ratio to allow the comfort of completely no clipping..
Digital recording seem to copy this strategy as it has a red area as well. So to me the sound level seems ok if it sometimes hits the red area.
And, i'm used to limit the play-back volume to max 70%. The remainder seems reserved to compensate for accidentely faint recordings.

Please tell me if i'm on the right track.

Nope!

The digital limit, digital full scale (0dBFS), does not have any room above it. Any bigger numbers are clipped. See the graphs in:

http://www.david.robinson.org/commsbill/#2_1_2


especially figure 2.1(d) clipped waveform.



If you've found a digital meter that shows 0dB, and lots of room above it, then 0dB on this meter does not equal 0dB FS. It probably equals -18dB FS or something. This means, on that scale, you're allowed an extra 18dB above the level marked 0dB before the signal clips.


You're right about an mp3 not clipping (due to an increased gain) until decoding and playback. With a suitable decoder, one that lets you get at the "unclipped" waveform (i.e. has a floating-point digital output, allowing numbers greater than 0dB FS) you can decode the thing, and them reduce the level, before sending it to the DAC. A DAC will always clip at 0dB FS because, by definition, that's the highest level it supports.


In Replay Gain, 89dB is an averaged perceived loudness, when replayed via an SMPTE RP200 calibrated audio system. It doesn't tell you anything about the maximum signal in the file. However, a 0dB FS sine wave at 2kHz has a replay gain perceived loudness of ~ 103dB. Other frequencies will give different results, because ReplayGain tries to take account of the different sensitivity of the human ear to different frequencies.


Cheers,
David.

This post has been edited by 2Bdecided: Aug 18 2003, 13:50
Go to the top of the page
+Quote Post
Joop
post Aug 18 2003, 19:29
Post #18





Group: Members
Posts: 4
Joined: 14-August 03
Member No.: 8366



Thanks a lot you all its completely clear to me now!
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: 27th December 2014 - 00:57