Hydrogenaudio Forums > Encspot and Winamp 2.77 both inaccurate for Bitrate

Dibrom

Nov 17 2001, 06:34

Since I've been doing lots of testing for the upcoming alt-presets I've discovered that both Encspot 1.0 and Winamp 2.77 are sometimes significantly inaccurate in the bitrate they list for a given file.

Some examples:

Castanets

LAME (209kbps):

CODE

C:audiowavs>lame --alt-preset normal castanets.wav castanets.mp3

LAME version 3.90 MMX (alpha 8, Nov 16 2001 20:25:26) ([url]http://www.mp3dev.org/[/url])

CPU features: i387, MMX (ASM used), SIMD, SIMD2

Using polyphase lowpass filter, transition band: 18671 Hz - 19205 Hz

Encoding castanets.wav to castanets.mp3

Encoding as 44.1 kHz VBR(q=4) j-stereo MPEG-1 Layer III (ca. 9.2x) qval=2

Frame | CPU time/estim | REAL time/estim | play/CPU | ETA

253/256 (99%)| 0:05/ 0:05| 0:05/ 0:05| 1.4053x| 0:00

32 [ 2] **

80 [ 6] ******

96 [ 0]

112 [ 1] *

128 [ 5] *****

160 [ 34] %****************************

192 [ 72] %************************************************************

224 [ 78] %*****************************************************************

256 [ 48] *****************************************

320 [ 10] *********

average: 209.1 kbps LR: 3 (1.172%) MS: 253 (98.83%)

Writing LAME Tag...done

Encspot (232kbps):

CODE

castanets.mp3

-------------

Bitrates:

----------------------------------------------------

32 | 0.8%

80 ||| 2.3%

112 0.4%

128 || 2.0%

160 ||||||||||||||||| 13.3%

192 |||||||||||||||||||||||||||||||||||| 28.1%

224 |||||||||||||||||||||||||||||||||||||||| 30.5%

256 |||||||||||||||||||||||| 18.8%

320 ||||| 3.9%

----------------------------------------------------

Type : mpeg 1 layer III

Bitrate : 233

Mode : joint stereo

Frequency : 44100 Hz

Frames : 256

Length : 00:00:06.0

Max Reservoir : 511

Emphasis : none

Scalefac : 13%

Complete : yes

Encoder : Lame 3.90

--[ EncSpot 1.0 ]--[ [url]http://www.guerillasoft.com[/url] ]--

Winamp (233kbps):

CODE

Size: 174999 bytes

Header found at: 0 bytes

Length: 6 seconds

MPEG 1.0 layer 3

233kbit (VBR), 256 frames

44100Hz Joint Stereo

CRCs: No

Copyrighted: No

Original: Yes

Emphasis: None

-

Bassdrum

LAME (196.6kbps):

CODE

C:audiowavs>lame --alt-preset normal bassdrum.wav bassdrum.mp3

LAME version 3.90 MMX (alpha 8, Nov 16 2001 20:25:26) ([url]http://www.mp3dev.org/[/url])

CPU features: i387, MMX (ASM used), SIMD, SIMD2

Using polyphase lowpass filter, transition band: 18671 Hz - 19205 Hz

Encoding bassdrum.wav to bassdrum.mp3

Encoding as 44.1 kHz VBR(q=4) j-stereo MPEG-1 Layer III (ca. 9.2x) qval=2

Frame | CPU time/estim | REAL time/estim | play/CPU | ETA

223/226 (99%)| 0:03/ 0:03| 0:03/ 0:03| 1.9727x| 0:00

32 [ 1] *

80 [ 3] ***

96 [ 3] ***

112 [ 6] *****

128 [ 7] ******

160 [ 36] %%%%%%%%%%%%%%%%%%%%%*****

192 [ 92] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%***************

224 [ 56] %%%%%%%%%%%%%****************************

256 [ 11] %%%%%***

320 [ 11] %%%%%%%*

average: 196.6 kbps LR: 131 (57.96%) MS: 95 (42.04%)

Writing LAME Tag...done

Encspot (232kbps):

CODE

bassdrum.mp3

------------

Bitrates:

----------------------------------------------------

32 0.4%

80 | 1.3%

96 | 1.3%

112 || 2.7%

128 ||| 3.1%

160 ||||||||||||||| 15.9%

192 |||||||||||||||||||||||||||||||||||||||| 40.7%

224 |||||||||||||||||||||||| 24.8%

256 |||| 4.9%

320 |||| 4.9%

----------------------------------------------------

Type : mpeg 1 layer III

Bitrate : 232

Mode : joint stereo

Frequency : 44100 Hz

Frames : 226

Length : 00:00:05.0

Max Reservoir : 511

Emphasis : none

Scalefac : 1%

Complete : yes

Encoder : Lame 3.90

--[ EncSpot 1.0 ]--[ [url]http://www.guerillasoft.com[/url] ]--

Winamp (233kbps):

CODE

Size: 145341 bytes

Header found at: 0 bytes

Length: 5 seconds

MPEG 1.0 layer 3

232kbit (VBR), 226 frames

44100Hz Joint Stereo

CRCs: No

Copyrighted: No

Original: Yes

Emphasis: None

-

So as you can see in both of these cases Encspot and Winamp report a significantly higher, and incorrect bitrate. I'd ask people to be aware of this when they post bitrates of various clips or when trying to gauge average bitrates of albums. Both of these programs seem to have flawed bitrate calculation and should not be used for such purposes until they are fixed. The only thing you should trust apparently is LAME as far as the correct bitrate goes.

Amadeus93

Nov 17 2001, 06:52

Well, if WinAmp is reporting the size correctly, then both it and EncSpot are right, and Lame itself is reporting incorrectly.

Castanets is 174999 bytes. (174999*8)/1000 = 1399.992 kb
Divide that by the number of seconds (6), and you get 233.332

To get Lame's result of 209.1, the sample would either have to be 156825 bytes in size or 6.7 seconds long.

I get the same results for Bassdrum.

JohnV

Nov 17 2001, 07:10

Castanets encoded and decoded with Lame is 6.662 seconds long. The original castanets.wav is 6.634.

Anyway 174999/6.662=26268.237 which is 26.268237 KB/s
and this times 8 is 210.14kbps.

So at least Lame reporting 209.1kbps is much more correct than 233kbps!

I believe Winamp shows this incorrectly because castanets is so short sample...

Volcano

Nov 17 2001, 10:55

Hmm. So, IF Winamp and EncSpot are both inaccurate, how can you get the correct bitrate then? I mean, if the file is ready encoded, and you didn't read the LAME output?

Speek

Nov 17 2001, 17:54

The MAD Winamp plug-in reports the bitrate correctly on Castanets.

Volcano

Nov 17 2001, 17:58

... and that applies to ID3v2-tagged files as well?

Speek

Nov 17 2001, 18:51

QUOTE

... and that applies to ID3v2-tagged files as well?

Don't know, I never ID-tag my files.

RD

Nov 17 2001, 21:38

Tag and Rename 2.0 reports bitrates that agree with the Mad mp3 plugin for winamp 2.77

Often they completely agree, and often they differ by 1 kbps (e.g., 202 vs. 203)

Tag and Rename 2.0 is also an excellent piece of software for, surprise surprise tagging and renaming....

Homepage:
http://www.softpointer.com/tr.htm

Amadeus93

Nov 17 2001, 22:58

Is the discrepancy between the numbers due to rounding? In other words, does Lame give an accurate rate because it is (for instance) using a time value of 6.662 seconds, whereas WinAmp and EncSpot just use 6 seconds?

robert

Nov 18 2001, 13:50

to calculate the correct average bitrate it is not important to know the exact duration of one frame. the only thing you need to know is, the number of frames together with the sumarized bitrates. The problem may be an inaccurate value for the number of frames in the Xing tag, as I believe that Winamp and Encsport rely on that information. This inaccuracy may have crept in due to the latest changes in VBR tag "improvements".