Sign in

username:

password:



Not a member?

Search speechcoding



Search tips

Subscribe to speechcoding



speechcoding by Keywords

ACELP | ADPCM | AMBE | AMR | AMR-NB | CELP | Codebook | DTMF | G.723 | G.726 | G.729 | GSM | Interpolation | LPC | LSF | LSP | MELP | PCM | Perceptual | Pitch | PSOL | QCELP | Quantization | SMV | VAD | Vocoder

Sponsor

NEW! TMS320C6474: 3x the performance. 1/3 the cost. Three 1 GHz cores on 1 chip.

Discussion Groups

Discussion Groups | Speech Coding | RE: conversion

Technical discussions related to Speech Coding (all itu and other vocoders, ACELP, CELP, AMR, etc)

  

Post a new Thread

RE: conversion - David M. Jack - May 29 1:57:00 2003



In order to do this efficiently you need to perform the conversion in
multiuple stages

See here :

http://www.soundslogical.com/support/resample/documentation/english/docu
mentparts/resamplehelp-30.html

It is also possible to combine the Interpolation and decimation filters
together by using a moving tap filter, this will reduce delay, and
provide a more memory efficient solution. This topic is covered in many
textbooks.

Dave > -----Original Message-----
> From:
> [mailto:]
> Sent: Thursday, 29 May 2003 7:22 AM
> To:
> Subject: [speechcoding] Digest Number 272 > ------------------------ Yahoo! Groups Sponsor
> ---------------------~--> Get A Free Psychic Reading! Your
> Online Answer To Life's Important Questions.
> http://us.click.yahoo.com/Lj3uPC/Me7FAA/CNxFAA/GP4qlB/TM
> --------------------------------------------------------------
> -------~->
>
> _____________________________________ > --------------------------------------------------------------
> ----------
>
> There is 1 message in this issue.
>
> Topics in this digest:
>
> 1. RE: [audiodsp] conversion
> From: > ______________________________________________________________
> __________
> ______________________________________________________________
> __________
>
> Message: 1
> Date: Wed, 28 May 2003 17:07:33 +0200
> From:
> Subject: RE: [audiodsp] conversion
>
> The answer from harishankar is not quite correct. Theory tells the
> following:
>
> Re-sampling from 44100 Hz to 48000 Hz:
>
> 1) Up-sample by expansion factor of 160 via zero insertion
> (sampling frequency
> is now 7.056 MHz!!)
>
> 2) Run up-sampled signal through a low-pass filter with
> cutoff 22050 Hz. This
> filter combines the interpolation filter required by the
> up-sampling
> process and the anti-aliasing filter for the following
> down-sampling process.
> Pass-band gain of this filter must be equal to the
> expansion factor of 160.
>
> 3) Down-sample filter output by compression factor of 147
> (sampling frequency
> is now 48000 Hz). > Re-sampling from 48000 Hz to 44100 Hz:
>
> 1) Up-sample by expansion factor of 147 via zero insertion
> (sampling frequency
> is now 7.056 MHz!!)
>
> 2) Run up-sampled signal through a low-pass filter with
> cutoff 22050 Hz. This
> filter combines the interpolation filter required by the
> up-sampling
> process and the anti-aliasing filter for the following
> down-sampling process.
> Pass-band gain of this filter must be equal to the
> expansion factor of 147.
>
> 3) Down-sample filter output by compression factor of 160
> (sampling frequency
> is now 44100 Hz). > The problem with this theoretical approach is the high
> intermediate sampling frequency of 7.056 MHz at which the
> low-pass filter needs to run. If we assume that this filter
> would need 100 cycles per sample, required processing power
> just for running the filter alone would reach 700 MIPS!!
>
> So I would think that there must be a more practical approach
> for this particular sampling frequency conversion. Does anyone know? > Regards,
> Roland >
> -----Original Message-----
> From: Simon Magus [mailto:]
> Sent: Wednesday, May 28, 2003 7:52 AM
> To: harima; ;
> ; ;
> matlab;
> Subject: Re: [audiodsp] conversion >
> harima <> wrote:Hi there,
>
> How can we do CD(44.1kbps) <---> DAT(48kbps)
> sample rate conversion.
>
> Thanks in advance,
>
> ciao,
> harishankar _________________________________________________________
>
> There is a subtopic on DSP called multirate signal
> Processing. From 44.1k to 48kHz you need to multiply by
> 480/441. That means you have to upsample by 480 and then
> downsample by 441. Upsampling is done by inserting 479 zeros
> in between the samples and the a low pass filtering with
> cutoff at 44100*480Hz. Downsampling is done by low pass
> filtering with cutoff at 48kHz and then removing 440 samples
> for every 441 samples
>
> Yahoo! Groups Sponsor
> _____________________________________ >
> ______________________________________________________________
> __________
> ______________________________________________________________
> __________




(You need to be a member of speechcoding -- send a blank email to speechcoding-subscribe@yahoogroups.com )