Subject: Re: [linux-audio-dev] Ardour : features page
From: Paul Barton-Davis (pbd_AT_Op.Net)
Date: Sun Oct 15 2000 - 06:53:34 EEST
Tom writes:
>I have questions (wishes) about Auto Loop.
>
>1) Does Auto Loop have a loop record mode? This would result in each
>pass through the loop being a new take. If so, does Ardour create a new
>file for each pass through the loop or does it append the new take to
>the end of the existing file? Is there the option of having the
>previous take automatically muted, such as when recording multiple takes
>of a single solo, or allowing all takes to play, such as when composing
>in layers.
I considered (and am still considering) this feature since I read
about it in Samplitude or some similar program. No, right now Ardour
does not do this.
I am not even sure this is possible with Ardour's model of a "take"
being its own file, and each take being sized to the length of the
whole tape (BUT, and this is important, NOT occupying that much disk
space - this is a standard POSIX trick). You would be creating one
file for every iteration of the loop. File creation is potentially
expensive, and the file size is large. Just to repeat: Ardour sizes
its files when they are created, but they don't actually consume disk
blocks until they need to do so. Suppose the loop was short: why on
earth would you want to create 20 copies of the recording ?
But even more fundamental it that I think it is ill-considered for
what Ardour represents. You are clearly thinking about a system like
the Jamman or the Echoplex. But those devices are not multichannel,
high-resolution HDR systems, they are simple two-channel recorders
with a very limited loop time. Ardour is emphatically NOT a
compositional tool, but should be considered a replacement for
something like an ADAT recorder, Mackie HDR (not yet available to be
replaced :) and so on.
I don't understand any situation where you would want to use a
multichannel recorder in this way. If you explain the
performance/composition context, maybe I would reconsider.
>2) Is dynamic looping available? By this I mean, can the end points be
>defined on the fly? This is only relevant if there is a loop record
>mode. I have not seen this feature in any music software, MIDI or
>digital audio, but I think it does exist in dedicated delay processors
>like the JamMan.
Having said all that, yes, dynamic (playback) looping is available and
works quite adequately for tracking situations. The only restriction
right now is that the step for incrementing and decrementing the loop
point is too large (1 second) because it is tied into the location
interface, which I considered to be for large-scale motion around the
recording. But I plan to change this to allow (possibly) single-sample
shifting of the loop points.
Keep in mind, however, that until Ardour has a tape butler that
understands the idea of an "advance directive" (i.e. don't get the
next N bytes from the file, figure out *which* N bytes we need next,
and get them), loops that exceed the amount stored in Ardour's buffers
will never operate seamlessly with a multichannel recording, because
it will take 10-500msec to refill the buffers at the loop return
point. The time will be hard to make totally determinate, which will
mess up tempo, and there will be a click every time.
>predefined before allowing the loop to function. Loop recording is
>great for improvisational composing, but how improvisational is it if
>the player must predefine the loop length, or must stop playing after
Right, I agree with this entirely, but you're thinking about a totally
different process. Let me put it this way: if I put you into a studio
with any recording system you care to name (ProTools, Logic, ADATs,
Ardour, whatever), would you try to use this program or hardwar to do
loop-based composition ? Or would you reach for your jamman, and
record what you did with it ? I know what I would do :)
>computer operator. I saw a video of Jaco Pastorius performing a solo
>concert this way. It was amazing. I think he was using a Lexicon
>JamMan, and I know other jazz players who do the same thing. Since
>Ardour has Dynamic Punch, it seems that allowing the punch points to
>also function as the loop points would take care of it.
Robert Fripp has been one of the prime exponents of this technique,
and what I have seen him do could not be accomplished with a
multichannel HDR system for quite some time to come.
But yes, the punch points can also be loop points, except that Auto
Loop is disabled during recording at this time.
>Feature 2 was the ability to play multiple sequences simultaneously and
>in sync while preserving the original time signature and tempo of the
>individual sequences.
Yes, you're right that this has nothing to do with Ardour. But just as
a side note, your explanation of why this matters to you and why the
usual note divisions of the western classical tradition don't work for
this: this is precisely why my program "SoftWerk", a pattern-oriented
MIDI sequencer that is more like the Doepfer Schaltwerk than Cubase,
completely avoids this notion of sequence tempo, and allows an
unlimited number of sequences each with an unlimited number of steps
to run at their own speed. I did this because I have a strong interest
in Indian rythmic patterns, and believe me, Cubase and the rest would
have a very hard time with a 9-1/2 beat cycle that was coupled to a
7-1/2 beat cycle and a 16 beat cycle all at the same time. Softwerk
can handle this with ease, but at the cost of giving up all that
"quarter note = ..." stuff that so many people are comfortable with.
>ps. I recommend removing reference to tape in your feature
>descriptions, since tape is never actually rolling. For example;
>
>Dynamic Punch
>
>When enabled, Dynamic Punch allows you punch in tracks on the fly.
>
>In a users manual it is appropriate to use tape as an analogy, from an
>educational perspective, to describe a particular operation, but in a
>simple definition of features or functions, don't refer to tape where
>there is no tape.
Excellent advice. I will try to do that.
--p
This archive was generated by hypermail 2b28 : Sun Oct 15 2000 - 07:20:30 EEST