On Thu, Jun 15, 2000 at 01:20:23AM -0700, Don Melton wrote:
> On Wed, Jun 14, 2000 at 11:45:46AM -0600, Mark Taylor wrote:
> > > Pierre Hugonnet wrote:
> > > Isn't this "Safe VBR" (let's say with 128kbs as a base bitrate)
> > > equivalent to a classical VBR with a minimum bitrate: lame -v -b 128
> >
> > The is one key difference: safe VBR mode chooses the total number of
> > bits to use before *any* quantization occurs, based on a simple
> > formula (also used by the CBR algorithm). The psycho acoustics are
> > only used in allocating the bits among the frequency bands.
> >
> > The other VBR modes decide on the total number of bits to use based
> > soley on how the quantization errors compare to the psycho acoustic
> > masking. Ideally, this is a better way to do things, if your psycho
> > acoustic model was perfect.
> >
> > But the psycho acoustic model in LAME far from perfect, and trusting
> > it as much as VBR does can lead to mistakes: A good example of this is
> > "vbrtest.wav", which was just added to
> > www.sulaco.org/mp3/gpsycho/quality.html
>
> Wow, that really _is_ hideous! Even worse than a "twangy" guitar sample
> I found the other day.
>
> That made me curious so I decided to do some quick comparisons. Here's
> some frivolous data on various encodings of "vbrtest.wav":
>
> Avg.
> Encoder options: bitrate Totally subjective comments
> ----------------------------------------------------------------------
> lame -v 111 Warble city! 31% of frames at 80 kbps
> lame -h 128 ;-) Sounds great!
> lame -h --abr 128 130 ABR indistinguishable from CBR
> xingmp3enc -V50 140 Sounds very good, but not "bright"
OK, I just pulled and built the latest CVS tree (mine was almost a week
old) and it looks like Mark's change earlier today to lame.c to disable
scalefac_scale for VBR modes has improved the "lame -v" case. It's
slightly better, the average bitrate has went up to 117, and only 22% of
the frames are at 80 kbps. Thanks Mark!
> Normally "lame -v" will produce files of higher average bitrates than
> "lame -h --abr 128" or "xingmp3enc -V50". Just how common is this kind
> of audio signal (he asks, shuddering)?
>
> I'm really curious about the Xing encoder's bit allocation scheme.
> Although their encoder gets a little "dirty" at the low and high end,
> they seem to really understand VBR. What are they doing right?
>
> Maybe "unsafe" VBR in LAME should strive to be more like the Xing
> results (without the distortion, of course) than like "safe" VBR.
>
> Also, I'm impressed with how good LAME is with a plain ol' 128 kbps CBR
> encoding. That kind of compression is the whole point of MP3, IMO.
--
Don Melton
mailto:[EMAIL PROTECTED]
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
