Hi,
you right about input format, output must be same, interim may be any.
Only by end you'll do output value=(short)(interim value+0.5);
You may trust to me nobody can to hear the harmony errors.
cheers

> We see that if we take the audio file. then the value is stored into
> the short int format. Now as in my case I have a 16 bit data. as an
> input.
> Now  what i want to know that the how the values which is say like
> 26.89 or 74.54
> etc are stored into the audio file..... as if we look the audio file
> then its all the integer...... now where are these datas of the
> fraction part.......
> 
>    please let me know about this
> 
> Thanks in advance
> Ranjeet

ranjeet.gupta@gmail.com (ranjeet) wrote in message news:<77c88a3b.0409010055.2a1d43b4@posting.google.com>...
> Hi all !!! 
> 
>   I am new to this field. I want to know in detail/brief that why we
> need to preprocess the speech signal. As i was going through the G.729
> annex B . In which it is written that if u take the input speech then
> u have to preprocess it before doing the manipulation/encoding it. So
> why we need to do this. ???
> 
> 
> We see that if we take the audio file. then the value is stored into
> the short int format. Now as in my case I have a 16 bit data. as an
> input.
> 
> Now  what i want to know that the how the values which is say like
> 26.89 or 74.54
> etc are stored into the audio file..... as if we look the audio file
> then its all the integer...... now where are these datas of the
> fraction part.......
> 
>    please let me know about this
> 
> Thanks in advance
> Ranjeet

   Preprocessing is done is to block the DC components(If any) in the signal.
   This prprocessing is nothing but a high pass filter with a cutoff near to    
   100Hz or so.

   Regarding representation of numbers they will be done using Q-format which
   is well known menthodology.

Hi all !!! 

  I am new to this field. I want to know in detail/brief that why we
need to preprocess the speech signal. As i was going through the G.729
annex B . In which it is written that if u take the input speech then
u have to preprocess it before doing the manipulation/encoding it. So
why we need to do this. ???


We see that if we take the audio file. then the value is stored into
the short int format. Now as in my case I have a 16 bit data. as an
input.

Now  what i want to know that the how the values which is say like
26.89 or 74.54
etc are stored into the audio file..... as if we look the audio file
then its all the integer...... now where are these datas of the
fraction part.......

   please let me know about this

Thanks in advance
Ranjeet