[Speex-dev] difference with NB and WB

Tom Grandgent tgrand at canvaslink.com
Thu Jul 28 05:39:27 PDT 2005


Although you can give Speex data at any sampling rate with any mode, 
it's a good idea to use the right mode for whatever rate you have.  
Speex is built for speech (not just any audio in general) and is 
sensitive to what kind of stuff is going on in certain frequency 
ranges.  If you tell it your data is at one sampling rate and give 
it data with another sampling rate, its idea of what frequency ranges 
are what will be all wrong, changing all the time, and it will be a 
real mess.  It's much better if you use the right mode for the data 
you have.  So if you have 8kHz, use NB.  If you have 16kHz, use WB.

(This sounds like a good question for the FAQ perhaps...  Hopefully 
I have the right idea with this explanation...?)


"Ronaldo Madrid" <Ronaldo.Madrid at asia.bigfoot.com> wrote:
> hi,
> i'm just a novice with a question.
> in the documentation, it says that "Speex is mainly designed for 3 different sampling rates: 8 kHz, 16 kHz, and 32 kHz.
> These are respectively refered to as narrowband, wideband and ultra-wideband."
> is this directly related to the sampling rate of the source sound. i.e. if i have 'PCM,8KHz,16bit,Mono', does it follow that i use NB for encoding/decoding? and if i have 'PCM,16KHz,16bit,Mono' do i use WB?
> although i obsereved that if i use only NB it doesn't matter what sampling rate the PCM is speex can still encode and decode it; and i can control the ouput quality with CBR.
> when will it be necessary to use WB? 
> thanks,
> ronald

More information about the Speex-dev mailing list