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: Short 3.96.1 Bit Rate Analysis (Read 3100 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Short 3.96.1 Bit Rate Analysis

I chose the song "Ocean Breathes Salty" by Modest Mouse becuase it suffers from bit rate bloat using V 2 (preset standard). A difficult song for LAME to encode. The purpose is to see how the V switches reduce the file size as they increase in compression.

-V 2 produces a file of 220 kbps, which is significantly higher than the targeted V 2 average of 192

-V 3 produces only a slight increase in compression, down to 193 kbps. Still higher than the targeted V2 average

-V 4 almost nothing is gained from increasing the compression from V3 to V4. The file size drops only to 190 kbps

-V 5 --athaa-senstivity 1 finally offers a significant gain in compression, down to 144 kbps. However, even this is still about 10% higher in file size than the targeted file size at V5 of 130

Conclusion: on this sample, there is no benefit from using V4 over V3 and little to using V4 or V3 over V2 (assuming compression size is a signifcant goal).

If others have done this type of analysis, I would be interested to see the results.

Short 3.96.1 Bit Rate Analysis

Reply #1 –
Quote
I chose the song "Ocean Breathes Salty" by Modest Mouse becuase it suffers from bit rate bloat using V 2 (preset standard). A difficult song for LAME to encode. The purpose is to see how the V switches reduce the file size as they increase in compression.


The bloat might be from the mp3 format's scalefactor band 21 problem. Rock genres in particular seem to suffer alot from this. If there is alot of high freq content then to encode it the mp3 rate will spike up.  The best way to control this artifical bitrate jump is the switch "-Y". Keep in mind its a little more complex than a simple lowpass. You can read about it all over this forum and the HA.org FAQ.

Try redoing the encodes with that switch and see if you get closer to the targeted bitrates.  ;)

Short 3.96.1 Bit Rate Analysis

Reply #2 –
V3-V9 already incorporate the -Y switch, so they already do as much as they can to solve the sfb21 bloat issue. Higher than average bitrate is also often caused by high stereo separation of the music, e.g. old Beatles songs. These are not particularly difficult to compress, but due to the sometimes extreme stereo separation Lame can hardly ever use MS frames, which increases the bitrate a lot.
BTW, I also didn't notice a significant bitrate drop from V3 to V4 on most music, but quite a drop when going to V5.

Greetings, sTisTi
Proverb for Paranoids: "If they can get you asking the wrong questions, they don't have to worry about answers."
-T. Pynchon (Gravity's Rainbow)
 

Short 3.96.1 Bit Rate Analysis

Reply #3 –
It's the quantisation measurement mode (-X switch).

For -V4 and higher bitrate, LAME uses -X 3,3
For -V5 and lower bitrate, LAME uses -X 1,0

This is why there is such a pronounced jump between 4 and 5.

Mode 3 is used because it is often an easy fix for a variety of sonic problems (it has stringent ways of measuring noise) and V4 and higher bitrate are quality-oriented presets. Unfortunately, files created with mode 3 measurement often suffer from dramatic bit bloat.