🚀 go-pugleaf

In article <1992Aug22.190716.2600@athena.mit.edu> parasite@athena.mit.edu (Vince) writes:
>
>I have a set of data which is mu-compressed from
>12 bits to 8 bits.  I need to maniulate it in
>its 12-bit form and I'm not sure what the
>conversion process between the two is.  All of
>the books I've looked in for an explanation have
>a few graphs displaying the wonderful properties
>of mu-encoding but no direct explanation of how to
>go from one to another.
>
>Does anyone out there know the definitions of the 
>encoding algorithm and wish to tell me what it is?
>If it makes a difference, this data was taken off 
>of the audio port of a Sun Sparcstation (/dev/audio).
>

mu-law is a code developed by Bell Labs about 30 years ago
for use in PCM telephony. It compresses a 13 bit dynamic
range into an 8 bit pseudo logarithmic form. The MSB is the sign
the next 3 bits are a scale factor and the least 4 are the
magnitude.  The spec is defined in various Bell & ATT documents
as well as CCITT Rec. G.711 and probably countless text books.

To paraphrase the table in CCITT G711

Segment  # of intervals     value at               code
                         segment end points
                           8159                  10000000
   8      16 * 256                               1000xxxx
                           4063                             
   7      16 * 128                               1001xxxx
                           2015
   6      16 * 64         
                            991
   5      16 * 32
                            479
   4      16 * 16
                            223
   3      16 * 8
                             95
   2      16 * 4
                             31
   1      15 * 2
           1 * 1
                              0                   11111111

This just shows the positive side - screwy isn't it?

Mike Gingell, Alcatel, Raleigh, NC USA  (919) 850-6444
UUCP:       ...!mcnc!aurgate!aurfs1!gingell
Internet:   gingell%aurfs1%aurgate@mcnc.org