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: Encoding times at different VBR and CBR settings (Read 11342 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Encoding times at different VBR and CBR settings

Just out of curiosity...

Which VBR settings are fastest when encoding using LAME? Would it be those at the lower end of bitrate of the spectrum such as -V6, or those of higher quality and less compression such as -V0?  On the one hand I'm thinking that less compression may mean less calculating to crunch things down, but on the other hand that to achieve the higher quality it may take more.

Where does 320 CBR encoding speed fall in comparison?

Encoding times at different VBR and CBR settings

Reply #1 –
Perhaps you could run some tests and find out?

Encoding times at different VBR and CBR settings

Reply #2 –
I could, but I expect that someone here has either already run the tests or else would know without even doing so.

Encoding times at different VBR and CBR settings

Reply #3 –
Quote from: JJZolx on
I could, but I expect that someone here has either already run the tests or else would know without even doing so.


Search and see if you can find something.

Encoding times at different VBR and CBR settings

Reply #4 –
I've been following HA since day one, and I don't believe I've seen your question asked or answered before. This is your opportunity to add some knowledge to the community.

Encoding times at different VBR and CBR settings

Reply #5 –
Ok, here are the results of a test I've run.  I tested the relative speed of encoding a small selection of my music library into Mp3 using LAME 3.98.4 at differenct VBR and CBR rates.  The files were originally FLAC encoded, decoded to WAV for this test, and consist of a fairly random mix of rock, jazz, blues, female vocals and classical, 73 files, 3.07 GB total size in WAV format.

The system on which the test was run consists of a dual core Intel Xeon E3120 @3.16 GHz, 4 GB RAM, running Windows Server 2003 SP2 32-bit.

The VBR encodings used only -V9, -V8, etc. as encoding parameters, and the CBR encodings used only "-b <bitrate>".  Files were tagged with some standard tags, ID2v2.3 only, using some generic strings, and output was to the same file name each time.  For example, at CBR 320 the command line used was:

lame.exe -b 320 --silent --id3v2-only --ta "Artist Name" --tt "Track Title" --tl "Album Name" --tn "1" --ty "2011" --tg "Genre" <input_file> "01 Test.mp3"

Code: [Select]
Encoding     Time     Seconds   MB/s
--------  ----------  -------  -----
   V9     0:04:39.84   279.84  11.23
   V8     0:06:22.67   382.67   8.22
   V7     0:06:40.14   400.14   7.86
   V6     0:06:47.97   407.97   7.71
   V5     0:06:58.54   418.54   7.51
   V4     0:07:06.17   426.17   7.38
   V3     0:07:10.96   430.96   7.29
   V2     0:07:19.61   439.61   7.15
   V1     0:07:25.66   445.66   7.05
   V0     0:07:32.88   452.88   6.94
CBR 64    0:06:53.63   413.63   7.60
CBR 96    0:09:35.23   575.23   5.47
CBR 128   0:10:48.50   648.50   4.85
CBR 192   0:09:12.16   552.16   5.69
CBR 256   0:09:23.64   563.64   5.58
CBR 320   0:09:28.69   568.69   5.53


Encoding at lower quality VBR settings is faster than higher quality, but the difference between each step is generally just a couple of percent.  Something like -V4 is only about 6% faster than -V0.

For the most part, the same trend was seen with CBR encoding, with encoding at higher bitrates being slower than lower bitrates.  There does seem to be something odd going on around CBR 128, which proved to be slower than anything else, including CBR 320. I can only assume something very different is going on in the encoding at that bitrate.

Encoding times at different VBR and CBR settings

Reply #6 –
Interesting! I wonder if someone can provide an explanation for the anomaly at 128 kbps (and perhaps 96 kbps). There’s a fair jump in speed at -V8 and -V9, too.

Encoding times at different VBR and CBR settings

Reply #7 –
Quote
There’s a fair jump in speed at -V8 and -V9, too.

Lame -V8 resamples input to 32kHz, and -V9 to 22.05 kHz.

Encoding times at different VBR and CBR settings

Reply #8 –
IIRC, the same happens at 96 and 64 kbps CBR. Why 128 is a bit slower than e.g. 192 kbps: no clue.

Chris
If I don't reply to your reply, it means I agree with you.

Encoding times at different VBR and CBR settings

Reply #9 –
I always thought that VBR would take longer to encode than CBR 
sin(α) = v sound/v object = Mach No.

Encoding times at different VBR and CBR settings

Reply #10 –
Quote from: SonicBooom! on
I always thought that VBR would take longer to encode than CBR 

Think of it this way. You have been asked to write a story. You have been told to adjust your writing style either toward being fairly wordy, or else more concise, so you keep this in mind while you write.

Now add the restriction that every paragraph must contain exactly the same number of words. Do you think that would make the job easier or more difficult?

Encoding times at different VBR and CBR settings

Reply #11 –
Quote from: pdq on
Quote from: SonicBooom! on
I always thought that VBR would take longer to encode than CBR 

Think of it this way. You have been asked to write a story. You have been told to adjust your writing style either toward being fairly wordy, or else more concise, so you keep this in mind while you write.

Now add the restriction that every paragraph must contain exactly the same number of words. Do you think that would make the job easier or more difficult?


Yeah, that makes sense now 
sin(α) = v sound/v object = Mach No.