Wideband quantization improved by adding a sub-frame gain, codebook

re-trained.


git-svn-id: http://svn.xiph.org/trunk/speex@3308 0101bb08-14d6-0310-b084-bc0e0c8e3800

1 files changed, 20 insertions, 2 deletions

diff --git a/README b/README
index b7578b0..73b4f2c 100644
--- a/README
+++ b/README
@@ -7,8 +7,26 @@ IP (VoIP). In some sense, it is meant to be complementary to the Ogg
 Vorbis codec.
 
 Although we aim at providing a patent-free codec, we strongly suggest you
-have a look at patent issues if you are thinking of using Speex 
-commercially. The speech coding field is a real patent mine field and
+have a look at patent issues if you are thinking about using Speex 
+commercially. The speech coding field is a real patent minefield and
 the scope and enforceability of all these patents is sometimes unclear,
 not to mention that each country has different laws.
 
+To use Speex:
+
+% speexenc [-n or -w] input_file.sw compressed_file.spx
+
+% speexdec compressed_file.spx output_file.sw
+
+where -n is for narrowband (8 kHz sampling) and -w is wideband (16 kHz 
+sampling).
+
+All audio files are assumed to be raw (no header) 16-bit PCM files. To
+convert a .wav file into this format:
+% sox my_file.wav -t sw my_file.sw
+
+Note that if the original sampling frequency of the wav is not 8 kHz or 16 kHz,
+you will need to add "-r 8000" or "-r 16000" (without the quotes).
+
+To create a .wav file:
+% sox -t sw my_file.sw my_file.wav