SndBuf: syntax for reading the number of samples

[laserbeak]

hi,
i'm new to chuck and find it very interesting. the fact that it works in real-time makes it very inspiring to experiment with code(i've been trying to learn c++ for 7 years and am still a newb for this reason)

anyway, to my question:
i dont know the syntax to get the samples from a sound in SndBuf.
let's say, for example, that i wanted to use C:\samples\drumloop

how would i type that correctly using the buf.samples command to find the number of samples?

-edit-
and are those number of samples in ms?

i've been trying a few things and i got this i think it works

[Kassen]

[laserbeak]

[kijjaz]

Laserbeak: It's nice seeing you here in chuck forum, my csound brother.. haha
ok anyway, let's play more with what you're playing.

now let's talk more about 'BPM'
we'll need a few mathematics.

ok.. let us have float bpm as the tempo of the music.
it's bpm beats per minute
= bpm beats per 60 seconds
= (bpm / 60) beats per second
so for one second, we've got (bpm/60) beats passed.

but what we're gonna use is the 'second passed.
so we need to calculate how long each beat takes.
1 second = bpm/60 beats
60 second = bpm beats
(60 / bpm) second = 1 beat

so actually, 1 beat equals 60/bpm
- - -

my idea is this:
// first, initiate BPM
180.0 => float BPM;
// calculate beat
60.0 / BPM :: second => dur beat;

// you can even calculate bar by define how many beats per bar
beat * 4 => dur bar;
- - -

with this, you can use beat or bar as the duration depending on the init value of BPM.
- - -

laserbeak: So does that mean that 58800 samples is 3000ms? I think that i would probably be easier to create a function or class that would convert samples to milliseconds for me to read from although math is math in any form.

kijjaz: you might have to check out Kassen's idea.
first, you'll need to find out the samplerate of the wavefile you're playing.

let's say we have SndBuf buf;
the sample rate of buf would be 'how many samples per second'
so that would be buf.samples() / buf.length() - - - (1)

then check out the current real-time sample rate
it's second / samp - - - (2)

to be able to play the wave with original speed,
we'll need to find the ratio between the sample's rate and the virtual machine's rate..
the ratio would be (1) / (2)

ok.. let's set the buf's read rate.
(buf.samples() / buf.length()) / (second / samp) => buf.rate;

and now you can play the file at the original rate without having to care about virtual machine's different sample rate.

!! ... this is my idea, still haven't been tested well
is my calculations correct?

[kijjaz]

i was wrong!

in chuck manual, SndBuf's length() fetches length in seconds
but actually ..

SndBuf buf => dac;
buf.read("special:glot_ahh");
<<< buf.length() >>>;

prints out: 992.000000 :(dur)

so this is one thing that needs to be changed in the manual.
now i can check out sample rate by..
<<< buf.samples() / (buf.length() / second) >>>;

prints out: 44100.000000 :(float) (works okay..)

[laserbeak]

where'd you find the .length() function????

[laserbeak]

so buf.length() / ms => float length; wil return the value in milliseconds. cool!!

thanks guys!!

[spencer]

[laserbeak]

[majutsu]

actually, even more interestingly, length() seems to return something very close to the number of samples despite the manual saying otherwise. but samples() returns a number of samples DIFFERENT than length(), in my experience a few thousand different. And neither matches the samples length as they were generated in me by soundforge or whatever. this makes beat slicing in chuck problem-ridden at this point. I either use single shot samples or slice the sample on the beat in chuck and play the slices forward or backward at different speeds. More beat-slicing accuracy than this in chuck is not had at this time (1.2.7b)

[laserbeak]

thats disapointing and you might be right if a bar at 180bpm is 1.5ms and not 1.3

i guess i could try slicing my loops and sequencing them that way

[kijjaz]

hmm.. is that so?
i didn't test .length() in many cases..
but i guess since there might be a problem with length,
we should stick to .samp() for now..

eh .. and is there any way to retrieve SampleRate without using .length() ?

[Kassen]