Forums

Best platforms (DSPs and/or FPGAs) to develop high end audio/voice applications on?

Started by Unknown September 8, 2007
Hi,
  This is a noob question - I'm doing some research in audio & voice,
and I need some advice.

What are the most popular platforms to implement high end audio &
voice applications on? To be more specific, I'm looking at audio/voice
applications that make heavy use of (large tap) digital filters, like
vocal harmonizers, digital mixing consoles, and acoustic echo
cancellers. What platforms do people develop on for these apps?

So far in my research, I've come across the TI 6713 DSP - a high
performance Floating Point DSP that is targeted for high end audio
applications.

1. What other choices are there in both the DSP and FPGA worlds?
2. Is it better to stick with DSPs, or are FPGAs popular as well for
these types of high-end audio applications?
3. And is Floating point generally used for these applications, or
fixed point?

thanks a bunch!

Huang

Hello,

just a thought, a standard PC with a quality soundcard may do...

An example program for full duplex IO here, just plug in your algorithm.
http://www.elisanet.fi/mnentwig/webroot/looprecord/index.html

Vladimir Vassilevsky has another full duplex code example for Windows on
his page that doesn't require portaudio.

Cheers

Markus 
huangpan@gmail.com wrote:
> Hi, > This is a noob question - I'm doing some research in audio & voice, > and I need some advice. > > What are the most popular platforms to implement high end audio & > voice applications on? To be more specific, I'm looking at audio/voice > applications that make heavy use of (large tap) digital filters, like > vocal harmonizers, digital mixing consoles, and acoustic echo > cancellers. What platforms do people develop on for these apps? > > So far in my research, I've come across the TI 6713 DSP - a high > performance Floating Point DSP that is targeted for high end audio > applications. > > 1. What other choices are there in both the DSP and FPGA worlds? > 2. Is it better to stick with DSPs, or are FPGAs popular as well for > these types of high-end audio applications? > 3. And is Floating point generally used for these applications, or > fixed point? > > thanks a bunch! > > Huang >
The "Digital Mixer" is in a different class from the other devices you list, not least becasue of the typically large I/O requirements, and a need for some modularity and scaleability. An example of a recent large-scale FPGA solution for a DAW is Fairlight "Crystal Core": http://www.fairlightau.com/default_content.html They claim to offer, thanks to the use of the FPGA, a dynamic adjustment between f/p and integer computation acccording to demands. ADI Sharc dsps are also widely used; e.g Studer (www.studer.ch) and Creamware (www.creamware.de) In contrast, ProTools still uses the 24bit fixed-point Motorola 56K series (www.digidesign.com), as do many smaller-scale hardware audio units. It seems one cannot easily generalise on the use of f/p v fixed-point, Fairlight for one cites the ability to adapt the hardware to divide the load between both modes, according to demand. I have heard of some hardware that uses TI dsps, but cannot recall any names offhand. Richard Dobson
great, thanks for the tips!