Using g729A enocder on Linux - strange results

[ramirose1]

Hello,
I had tried to use g729A enocder from
the ipp sample on Linux (ipp 4.1).

I am working on a Linux RedHat 9 machine.
I had build the g729 sample for IPP 4.1
on this Linux.

My input PCM file is a valid
16-bit 8 KHz ,signed,little endian,mono file.


Since I want to encode this PCM file into
729A format, I tried:
./g729iencoder -rA in1.snd in1.encoded.out

it does run; however, the size of
the input (in1.snd) is 116 k and the size of the
encoded output file (in1.encoded.out) is 119k !

It seems to me starnge as I expected the output file size to be
about tenth of the input.

Why is it so ? Am I doing something wrong?

Regards,
Rami

[Vyacheslav_Baranniko]

Hello, the g729 and other IPP 4.1 speech coding samples produces output bitstream file in test bitstream file format used in standard testing software. Generally, the in the test format each bit is stored as 16bit WORD signature: 0 - 0x007f, 1 - 0x0081. Plus a small header per frame as also stored. That means a bitstream is produced ~16 time longer than you might expect. Rgds Vyacheslav