hi,

i've been playing for a month or two now with pd and this really looks like the toy i've been looking for for a long time. i'd like to try it out on live gigs but i have a few questions. I'm currently on the WinXP 0.38.4 extended edition with a presonus firebox as audio device. So here are my questions :

• what about latency? currently it can't seem to get under 20/25ms because then portaudio starts crackling away. Would this be better under linux with jack, and is my soundcard supported under linux?
• how comes pd freezes (for a couple of milliseconds depending on the size) when loading a sample or a second patch? This is very annoying especially when playing live... are there any plans on improving this in the future or do we have to live with it?
• has anyone has live experience with pd - i'm sure you all have - so how do you overcome the problems mentioned above?

all advice is welcome.

cheers,

domien

I've never tried it personally, and someone please correct me if I'm in error, but apparently the best way to load a sample up live is to have a subpatch with a very high sampling rate and use readsf~ and tabwrite~. This way it will only try and load some number of samples per dsp block, instead of trying to load the entire file as quickly as possible, which can overload the cpu.

yeah, pd always cuts when loading sound files bigger than a few milliseconds.

you can try the method suggested above,
..or else make yourself a file loader, and put heaps of arrays in a subpatch, and load all the samples you need before you start, and then just use [set< messages to change between different arrays. ..because the samples are already stored in memory, there will be no audio cuts.

sorry, no answer about latency cos i've never worried about that,.

I do what hardoff does and load all my samples before I start, which gets annoying, so I do more synthesis stuff instead. The same with video stuff, I load all the images before I start, because otherwise there are unpredictable delays at inconvenient times...

I don't worry too much about latency, but then I'm not processing other people's sounds, and I play using step-sequencers mostly, so there's a layer between my actions and the sounds.

And when I do audio<->video stuff, I often use a latency of 200ms or more, because GridFlow does its processing in message-time, so all the video processing has to be complete before the audio can carry on computing.

Thanx to you all guys for the advice. I had been thinking to preload many samples, but then again this is somewhat limiting because you have to anticipate for everything, whereas when playing live you're just gonna need that sample that you didn't load...

On the other hand I still wonder if anyone uses pd to process live input? Does anyone know if there are plans to offer proper ASIO support (like Max, Reaktor,...)?

Cheers,

Domien