I have managed to create something that takes a sample and can play it forwards or backwards, can loop forwards or backwards, and can change the speed.....

What I am wanting to do now is be able to record incoming audio and manipulate it this way. In other words, how do I create an SK-1 style sampler where I can hit a key and it will record the audio coming into the audio input, and then take that same little clip of audio and mess with it. As it is now I load samples in using soundfiler.

I guess I'm kinda trying to make that sample loopy scratchy time-stretch thing on Kaoss pads, and all I need now is to know how to grab little sampes of incoming audio.

Any ideas? Thanks.

If you want your samples to be a predetermined length (like four beats or whatever), you could just make a table/array at whatever size you need it, route the incoming audio to a [tabwrite~], and bang it when you want to record. This will probably be the easiest as I would imagine you're manipulating a table with a sample to begin with.

If the length of the sample is uncertain, you could use [writesf~], then reload the sample into a table using [soundfiler] with the -resize flag.

I'd prefer to press a key to start recording, then press a key to stop recording, and hopefully have that audio end up in the looper thing with as few steps in between as possible. I'm not really thinking about it in terms of pre-determined lengths. I just want to be able to do it on the fly.

With that being said, which would you recommend?

Also, I have no prior experience with using writesf~ so that's kinda over my head at this point.

you want to go with maelstorm's first suggestion then, i think. make a [tabwrite~] object, and right (ctrl) click on it to see its help file.

Well, since you're using looper, you probably already have a predetermined length just so the loops can stay in sync. So banging a [tabwrite~] will fill the table at whatever your set loop length is if the size of the table is that long. This will probably cause you less headaches. That's, of course, assuming your looper only uses one length or you're not automatically speeding up/slowing down files of varying lengths to sync them up. It kind of just depends on how complex your patch is.

But if you really want to just record and stop at whatever arbitrary place you decide, I think [writesf~] is the way to go. When you press the key to stop recording, you can have hit a [t b b] setup that will stop recording and open the recorded file into a table using [soundfiler]. That way, while there are multiple steps going on in Pd, on the user's end it's just record and stop.

Cool. Yeah, maybe my use of the term looper was confusing. The thing I made loads up a sample and takes the length of the sample and jumps back to the beginning once the end point is reached numerically, hence looping. But I'm not necessarily using it to loop musical phrases and syncing them up to anything.

I'll probably end up trying both methods, but I just wanted to know where to start. It looks like I have some learning to do with writesf~

Thanks everybody.

I think I figured it out!!! I ended up using writesf~ and I was able to hook it into my existing patch without removing any of the original capabilities. Now I can grab incoming audio and play it with my speed-up, slow-down, forwards and backwards player/looper!!!

Amazing. Thanks again... It's a mess (as are most of my patches) but it works the way I want it to. Maybe I'll post it once I do some fine-tuning.

> Maybe I'll post it once I do some fine-tuning.

Please do. It should be interesting.

The disadvantage of using writesf~ is that you're using the computers disk which is far slower than ram. Depending on the machine this might cause performance issues. One way to get around this on linux is to use RAMFS or TMPFS. It lets you use a chunk of ram as if it were a disk, allowing you to "save" file there.

http://en.wikipedia.org/wiki/TMPFS

There is a way to do this on OS X: http://lists.apple.com/archives/darwin-kernel/2004/Sep/msg00004.html as well.

That's a sweet tip, Saturno. Thanks!

Be careful with RAMFS and TMPFS though since their size is fixed and you can get a disk out of space error. Be extra careful with RAMFS since you could possible eat up your entire ram if your not careful. TMPFS is a little safer since it will use virtual memory when need be.

10MB of ram disk would only get you about 79 seconds of audio at 24bits/44.1kHz

You could go with 100MB for 792 seconds but that's 100MB of ram your machine wouldn't be able to use for anything else.

That's all way over my head. I'm using a Mac OSX and it seemed to work right off the bat with no problems. I'm pretty much only planning to use it for relatively short bits of audio (probablyl up to 20 seconds at the most) so far no glitches or anything like that. It works exactly how I wanted. I'm happy.

This more or less eliminates any desire to buy a Kaoss pad, which I had been thinking about for several years now. Thanks for saving me a few hundred bucks.

Back again... with a bit of bad news and another question.

Two days ago my hard drive crashed on my macbook. This is the 2nd time that has happened to my laptop, except this time I had mostly everything backed up. (Last time not so much..... goodbye vacation photos.... IDIOT!!!) Anyway, I lost all the stuff I had made in the last couple weeks, but I'm sure I can figure it all out again.

Now some of the posts on this thread talked about recording to the hard drive using writesf~ which is what I had been doing. My question is, is it possible that writing to the drive like this broke my hard drive?

(It's under warranty so I'm not really upset.... just wondering if I should not do that again when I get it back with a new drive.)

I never recorded more than 20-30 seconds of audio using this (or any other Pd method) so I doubt it. I'm guessing it just burnt out for some other reason. But I'd like to avoid this happening again if writesf~ had anything to do with it.

Any thoughts.

That would be rather strange i guess. I've been recording a lot of impro sessions during which i write 5 simultaneous stereo files (using 5 [writesf~] objects) on different computers/usb drives, never had any trouble with it.

Yeah, I've recorded 45 minutes straight a few times using [writesf~], and my hard drive still works. Which is good, because I'm a lazy bastard when it comes to backing up.

Well, this is my 2nd hard drive replacement on my macbook (which I've only had since August of 2006) so I don't know. Maybe I'm letting it get too hot, or maybe I rattled it loose by moving it too much while it is powered on. Who knows?

You know, most people I know who've dealt with MacBook hard drive failures (which is quite a few as my school requires students to have MacBook Pros) have had them die within a year or two, so they may just not have a great lifespan.

hmm...

...I should go back my stuff up...

You know, most people I know who've dealt with MacBook hard drive failures (which is quite a few as my school requires students to have MacBook Pros) have had them die within a year or two, so they may just not have a great lifespan.

Of the two people i know that have macbooks bought about 2 years ago, their hard drives have failed and been replaced. I remember the iBooks had cracked logic boards a few years back and loads had to be recalled. I heard that Apple trying to source cheap labour and parts. Things like too much thermal paste on the cpu leading to overheating and causeing all sorts of problems.

Touch wood my first generation 17" powerbook is still trundling along nicely, albeit slowly!

any how i digress.....

OK.... so back to my original question. Unfortunately I had everything all figured out and everything was working exactly how I wanted, until my hard drive broke and I lost it. (I didn't lose everything, just the part about recording audio into my scratchy sampler)

Anyway, I figured it'd be easy to redo it again, but I can't get it. So how exactly do I get from writesf~ to loading the recorded file into a table? I know that's how I had it set up, and I'm totally stumped. I need to get from audio coming into writesf~ and then read it out of a table.

Yet another hard lesson learned: always back your stuff up!!!

P.S. - I have been looking at the help files (which is how I figured it out the first time) but I'm just stuck. Maybe once I figure something out, the knowlege I needed to do it flies out of my head because I don't think I need it anymore. I feel dumb.

UPDATE!!!!! Nevermind. Just figured it out again. And saved it. And backed it up. And emailed it to myself.

So you're recording the sample onto your harddrive with a writesf~ and then loading it to a table?

Wouldn't it be easier to record the sample directly to the array?

My (limited) understanding was that if I wanted to be able to start and stop recording on the fly and then loop the recorded sample, I'd have to use writesf~ but yeah, if it's possible to do it directly to an array (of a non-predetermined length) then that probably would be better.

The whole issue I think is that I'm not starting out with a specific length of time for the sample being recorded. I'm just pressing a key to start and then pressing another key to stop.

Any suggestions on doing it better/more efficiently?