DSP for Audio
Yes, all of these DSPs are great for DSP applications. However, processing 16 bit (or even worse, 20/24bit) audio samples with a 16bit (or 24 bit) DSP will make it necessary that you use "emulated floating point" operation on our DSP, becasue of things like SNR, Noise Floor, Dynamic Range, etc.
That means, more difficulty for programming, because you (as the DSP programmer) will need to handle all the issues regarding fixed-point data formats. And the effective processing power of your DSP decreases rapidly. Anyway, for stablished algos like Param Eq and the like, there are known implementation structures which make it easier to be handled in such devices.
Having a DSP like the SHARCs (21065L or 21161N), on the other hand, gives you some advantages. 32 bit fixed point, 32 bit floating point or even 40 bit floating point. What else would you wish for audio processing in _real_ real time? They already incorporate some nice features: SIMD architecture (2116x), large on chip dual ported SRAM, serial audio interfaces like I2S, 14 DMA channels, on chip SDRAM controller, and many other which I can't remember at the moment. Go to (
www.analog.com/dsp).
About the EZKITs (I'll talk only about the one for 21161, the one for 21065L is a bit more modest, still _very_ powerful -> EMAFE interface)
- ADSP-21161N SHARC DSP
- 48 Mbit SDRAM
- AD1836 96 kHz Multichannel Audio CODEC
- AD1852 96 kHz auxiliary DAC
(Two stereo input channels, 8 stereo output channels)
- SPI Interface
- USB 1.1 interface
- JTAG Interface
- 4Mbit Flash
- VisualDSP++ SW Toolset
Maybe the price tag is a bit higher (US$ 400-500 range), but it is well worth the investment.
JaaC
www.sanjaac.com