[Tobiah[_6_]]

Or something quite similar. I would like to sample a guitar
chromatically over the range of the instrument with multiple velocity
layers. This will require the generation of hundreds of individual
sample files that I can't be bothered to curate by hand. My solution
is to record each sample in some logical order into one long sound
file, then to split, and fade in/out the samples automatically using
a command line program. It would take arguments to configure the
split, and generate logical names for the files that my sampler might
be able to consume with little manual assistance.

So I'm asking about detection of the start and end of the samples,
and after fade in/out schemes. My initial thoughts outline and
argument list something like this:

*) Threshold. An initial level that would trigger the start of a
sample. The sound file library I'm using (libsndfile) presents each
sample as a normalized float, so I thought I'd make this a straight
fraction of max, like .03. After cursory consideration, it seems
that a single sample over this limit would suffice as a trigger if
the recording environment is reasonably quiet. Maybe there is more
to consider?

*) Attack. A fade-in time that would be applied to the samples
preceding the threshold hit. I was thinking linear, but maybe
exponential or log or something makes more sense?

*) Hysteresis. A second (possibly lower) threshold value that would
trigger the end of the sample. In this case, I can see that I'd need
to designate a minimum length of time through which this threshold was
*not* hit by any sample.

*) Decay. Length of time over which to taper the tail of the sample
so that no clip is heard at the end.

Before I wast time going down wrong paths, I though I'd see whether
I could get any suggestions, or I suppose a pointer to a program
that already does this nicely.

[geoff]

On 29/04/2020 2:07 am, Tobiah wrote:


Before I wast time going down wrong paths, I though I'd see whether
I could get any suggestions, or I suppose a pointer to a program
that already does this nicely.


Sound Forge has very powerful scripting. But you'd still need to develop
the script (or customise an existing one) if you can't find one from
withing the user base.

geoff