DIY pocket POD

[patricks]

Hi all,

What are your thoughts on a stompbox designed to load amp/cab impulse responses? I know of the Line 6 POD/pocket POD (http://line6.com/pocketpod/) and the Torpedo CAB (http://www.two-notes.com/en/hardware/torpedo-cab/), but I'd like to know how feasible it would be to build a DIY version.

I'm not talking about re-creating either of them, both the pocket POD and the Torpedo have a huge range of options and I imagine they'd be very difficult and expensive to emulate. What I'm thinking of is something with maybe 5 preset slots and a USB port so that you could to load impulse responses from your computer to the pedal. Presets would be selected by either rotary encoder or stomp switches, but only one preset would be active at a time. Stereo ins and outs would be nice, but I'm not sure whether that'd require too much processing power.

Please forgive me if I'm talking rubbish, I have zero experience with digital effects
Thanks in advance!

[CynicalMan]

I don't think that would be possible with currently available DIY materials. The closest you could get would be the Spin FV-1, but IIRC it can't do convolution. I'd probably go with a PC and a midi footswitch. Although, you might be able to put something together with one of the mini-computers that have been becoming available recently (Raspberry Pi, VIA APC, etc.).

[patricks]

Thanks for the info, those mini-computers look really interesting! There's a discussion on the Raspberry Pi forums about using it for amp modeling: http://www.raspberrypi.org/phpBB3/viewtopic.php?f=41&t=945&p=13675&hilit=impulse+response#p13675
I think it could be done, but it'll have to be done by more capable people than I 

[meffcio]

I don't think such thing would be hard to do. You just need and ADC, DSP processor, DAC, and some kind of interface. It's certainly doable.
Look at this bastard:
http://howleraudio.com/frontpage/

[mhelin]

Recommend you to go to markseel's Simple DSP II thread. I think the XMOS device is capable to handle at least short impulses. A single thread has 61.5 MIPS performance, and if you partition the FIR between multiple threads and use moderate sampling rate (POD XT for an example has 39062.5 Hz sampling rate, POD/2.0 33 kHz, 32 kHz is anyway more than enough for guitar) I assume (@32 kHz rate) that single thread can do 300 point FIR so four threads could do about one thousand points FIR. If multirate FIR could be implemented on XMOS that would make some improvement. Anyway the impulses on net are from 50k to over 100k bytes in size and can't even be fit on XMOS processor's internal SRAM as such. For an example the GuitarHack impulses which though are bad because most are not trimmed and are so too big in size and consume too much CPU if brute force FIR is used (with FFT/iFFT the size doesn't matter as the impulse is FFT'd only once). I tried to trim one impulse "GuitarHack Fredman Straight -20.wav" which was 203 kb in size. I trimmed it, converted to mono and saved @32 kHz sample rate 16 bits. It took only 5 kbytes after that, but I'm not sure if XMOS can do even that short impulses (the FIR is about 2500 taps because 16-bit samples).

[patricks]

That Howler Audio box looks like the sort of thing I'm thinking of, thanks for the pointer
I'll check out markseel's Simple DSP II thread, too, thanks for that!