Basic Components
1) Soundfile window
Containing the names, attributes, and user-specified gains for up to
240 soundfiles.
2) Playnote Window
Containing playnote commands, telling the driver which segments
of which soundfiles to load, when and how.
3) Information Window
Containing reports from the sound driver. Check this for error messages
4) Tracks Window
Containing gain controls for each channel of each of the 12 tracks
and the 2 output tracks as well as reports on peak amplitudes and toggle
switches for each track.
5) Various Buttons
To load/reload data, play, stop, pause, and start over.
6) Information Fields
To specify the number of tracks, the starting and ending time, the
elapsed time, the timescale.
7) Track Map Window
Contains a visual layout of soundfile segments on tracks, and a marching
timeline while playing mixes
8) Safety Backup
The path names of the relevant soundfiles are listed in the Soundfiles
window. The number of the soundfile, listed immediately to its left, is
important and is used by playnote() to reference that soundfile.
Soundfiles can be loaded by using the Open Input Soundfile command
in the Soundfiles menu, or by entering the name in the form above
the window, and hitting return, or hitting the load soundfile button.
To load a soundfile at a specific location in the soundfiles window, preposition
the cursor on that line. If you want to load a placeholder at some point,
use the name 'null'.
Soundfiles can be played directly
from the soundfile window by simply double clicking on the name of the
soundfile. A second click anywhere in the window will stop the play.
Soundfiles can be turned off and on either
during or before a mix is played by clicking once on the soundfile line
in the on_off window
The gain column is editable. You can
change a gain by selecting and typing over, or inserting, or, you can place
the cursor in the gain and move the slider above the gain column to change
that gain. These gains control the amplitude of that soundfile, and are
multiplied by the gain factor on the playnote card, the track gain for
a given playnote and the output gain. To fine-tune a gain you can use the
arrow keys while the gain slider is selected.
In the playnote window you put the actual data for accessing
the soundfiles. Each playnote() command is usefully thought of as
a note. It instructs the sound driver to play part or all of a soundfile
at a given time, on a given track, with certain properties. You can copy
and paste text into the playnote window. The arguments can be placed in
any order. Both the playnote window and the large
playnote window use Emacs key binding. This is set in the resource
file.
Each playnote() command must appear on a separate line in the
playnote
window. It needn't begin in the first column.
The only arguments for a playnote
window are the track and the sound numbers (track, and snd). All other
arguments have default values if omitted.
Each note as specified by a playnote() command has the
following properties, specified with these names:
skip= : amount of time to skip into
that soundfile before beginning to read it. Default is skip=0.
at = : time in the output mix to start
playing that segment of the soundfile. Default, at = 0.
track= : the track number that soundfile
will occupy. Note that only one sound can be played on a track at a given
time. There is no default, the track number must be specified. If two playnotes
occupy the same track at the same time, or overlap for a period of time,
the one that was entered second will destroy the overlapping contents of
the first one. Note that this is important if you are using the reload
playnotes button. If there is no track specified the next available
track will be allocated. This can lead to confusion when notes are inserted,
etc., since the track will be one greater, (mod the number of tracks) than
the previous playnote. It is a good idea when using this feature to specify
the track for the first playnote. This is useful when using a bunch of
playnotes which have successively later starting times. Otherwise care
must be taken that segments do not overlap on the same track. See the trackrange()
command.
snd= : the actual number of the soundfile,
as referenced by its position in the soundfile window. There is no default,
the snd number must be specified.
dur= : amount of time to play that
segment. The Default is to the end of the soundfile. When the transp=
argument
is invoked, a positive dur value will indicate the amount of time to read
the input file, so that for downward transpositions the actual duration
will end up being longer, and for upward transpositions, shorter. If the
argument is a negative value it represents the absolute amount of time
to play that note, regardless of the transposition (e.g. it represents
the output rather than input time.)
end= : the time on the input file
to stop playing. i.e. end = skip + dur. It can be specified instead of
dur. The Default is the end of the soundfile. If transp is
in effect and the argument is positive it represents the end on the input
(as with the dur argument), and if negative it represents the end time
on output, relative to the starting time of the note.
transp=: this is a single valued
gliss function. e.g. transp = -2 is equivalent to gliss(0,-2,1,-2).
It will simply transpose the whole passage down by 2 semitones.
mtransp=: instead of indicating
transposition by semitones, this indicates it by multiplying the frequency
by the specified constant. e.g. mtransp=2. will raise the pitch
one octave, mtransp=.5 will lower it an octave etc.
amp(time,amp,time,amp etc): an envelope
for the note which will control the amplitude as it plays. The times are
normally relative to the length of that note (unless you use a stickpoint,
see
below) so any arbitrary timescale can be used. There is no maximum gain
and fractional values can be used. Default: all amps set to 1. Note that
you just specify numbers within the parentheses. e.g. amp(0,0, 1,.5,
2,0) will create an envelope which goes from 0 at the beginning of
the note, to .5 in the middle, to 0 at the end.
gliss(time,interval,time,interval etc):
this will cause the apparent sampling rate of the sound to change over
time. Note, however, that simple interpolation is used so spectral artifacts
will be created. Default: no gliss. (Note that transposition up
will increase the data transfer rates. ) The interval arguments are given
in semitones (with fractional parts allowed) up or down. For example,
gliss(0,1,1,2,2,-1.5)
will
cause the note to start one semitone higher than the input file, go up
to two semitones higher in the middle of the note and end 1.5 semitones
lower.
mgliss(time, freq.mult, time, freq.mult,
etc):, as with mtransp, the transposition arguments here are
expressed as multipliers of the frequency, rather than in semitons.
If you have both a gliss() and transp= argument, then
they will be added. All pitch values in the gliss() curve will have
the transp= value added to them.
ampl(time,amp,time,amp, etc): a separate
amplitude control for the left channel. Default: set to 1. Syntax and time
scaling are the same as for the amp() command.
ampr(): is similarly a separate control
for the right channel. Default, set to 1.
STICKPOINT: This feature allows
any of the above commands which have a string of time/value arguments to
use absolute time rather than relative time, for the beginning and end
of the notes. The stickpoint is a | sign. You specify it
in the middle of a set of arguments to "spread" or "shrink" the middle
of the note, while leaving the times around the stickpoint fixed.
amp(0,.2,2,.5, | , 9 1, 10, 0) for example, means that this note
will have a rise from .2 to .5 over the first 2 seconds, and have a decay
from 1 to 0 over the last 1 second of the note, (assuming its duration
is at least 3 seconds), and between these two times the amplitude will
go from .5 to 1. If, for example the note is 20 seconds long, this is equivalent
to saying amp(0,.2,2,.5,19,1,20,0).
If there were not a stickpoint in the above example and the note were
20 seconds long, it would equivalent to saying amp(0,.2,4,.5,18,1,20,0).
Here
you can see that the times are warped relative to the actual duration of
the note. There can be only one stickpoint in an argument set, but there
can be any number of argument pairs around it.
gain=: the signal is multiplied by
the product of this gain (default = 1) and the gain specified for that
track in the tracks window. It is up to you to make sure that the product
of any sample and the gains does not exceed +/- 32767. (You'll know if
it does.)
pan= : Normally a mono input file
will be split evenly between the two stereo output channels. A stereo input
file will be mapped channel by channel. If you specify the pan argument
this is equivalent to multiplying the gain of the left channel by the pan
value and on the right channel by 1.0 - the pan value. It is equivalent
to using ampl() and ampr(). In other words, for a mono signal
pan=.1
is
equivalent to shifting the signal so that it appears to be about 10% of
the way between the left and right channels. If you are using the
pan
feature
the amplitude of the signal will be boosted to accomodate the loss in power
a signal has when it is shared between two speakers. . The formula used
is 1./sqrt(pan*pan + (1-pan)*(1-pan)).
rev: This word by itself, (or reverse)
will
cause the segment to be played backwards. (Thanks to Pete Yadlowsky for
this hack.) In this case skip will indicate the amount of time to
skip backwards from the end of the note. (NOTE the rev function
is broken in this version, don't use, yet.)
playnote(snd=1,track=1,skip=2.5,at=3.5,amp(0,0,1,1,2,0),dur=4)
will cause snd 1 to be played on track 1, starting at 2.5 seconds into
the sound, starting at time 3.5 in the mix, for 4 seconds with an envelope
going from 0 to 1 and back down.
playnote(track=1,snd=1,transp=-2.5)
will simply play all of snd 1 on track 1 down 2.5 semitones.
Thecommands in the playnote
window are initially loaded by hitting the load everything, or
reload
playnotes buttons in the control window. If some of the text in the
playnote
window is selected, only that segment will be loaded. You can subsequently
select a subset of the playnote() commands in the
playnote window
to be added to the mix, or altered within the mix by simply selecting them
with the cursor and hitting the reoad playnotes
button. This will
save a bit of time since it sometimes takes a few seconds to load in a
bunch of files and playnote commands. Note that if a subsequently loaded
playnote command occupies the same time portion of a track as a previous
one, then the first one will be eliminated.
last=: This is a substitute for the
at=
command.
It will remember the ending time of the last
playnote()
and
begin this playnote at that time plus the value set by last=.
So,
for example, if you specify last=0, the current note will begin
exactly where the previous one left off (to within the current timescale),
or last=-1, will set the current note to begin 1 second before the
previous note ends. This is quite useful if you are doing splicing. You
can create a butt-end splice and then bevel and overlap without having
to do any arithmetic.
overlap=: This is the same as
the last= command except that it worries about the beginning of
the previous playnote() rather than the end. So overlap=0 will
start the current playnote() at the same time as the beginning of
the previous playnote().
When either of these arguments is used the actual computed at value
will appear in the output window. I recommend pasting this into your playnote,
replacing the last or overlap definitions, to avoid too much
confusion. Things can get messy, particularly in connection with the offset=
command
(see below).
Additional commands which
can appear in the Playnote window
These arguments are specified on separate lines, not within a playnote()
argument
list, and will effect all playnote() commands which follow.
This will result in the value x being added to the at= value
of any playnote() which follows this statement. This is a convenient
way to alter the starting time of contiguous groups of playnotes.
This will result in the value x being added to the currently loaded
offset
value.
This is a convenient way to increment or decrement currently defined offset
values.
It will only effect playnotes which follow it. If you are using this feature
it is highly recommended that the first line in the window be offset=0,
so that everytime you reload playnotes the cumulative
addoffset
will be the same.
trackgain(starttrack,endtrack, time1,amp1,
time2,amp2... timen,ampn)
This will put a global amplitude curve on all notes occupying the tracks
between and including starttrack and endtrack. Here the times
are absolute. For these tracks, any referenced times before time1 will
be assigned amp1, and after timen, will be assigned ampn.
The
default amplitude for all tracks is 1. For this feature it is best to use
amplitudes greater than 0 and less than or equal to 1. This will keep you
out of trouble with respect to amplitude overflow since these values will
be multiplied by the gain set on individual tracks in the track window,
the gain for each sound, and the amplitude settings for each playnote.
There can only be one trackgain() call for any track or subset of
tracks. (I'm working on a better system.) Note: you do not use names
in this call, just numbers.
trackrange(starting track, ending track)
This statement will cause all times subsequently referred to by playnote()
at= statements to be multiplied by a factor. If x is less than 15 it
is a simple multiplier. If greater than 15 the factor is formed by 60/x,
so standard tempo speeds can be used.
any text on any line which is preceded by // will be ignored by the
driver. e.g.
//this is a comment
playnote(track=1,snd=1)//this is another comment, but the playnote
is taken seriously
The third window is simply a larger version of the playnote window on
the main page. The purpose of this is to allow you to look at larger groups
of playnotes without having to scroll around. In addition, if you like
smaller, or larger fonts, you can change these by editing the resource
file. The name of this widget is 'playnote_bigtext'. When you toggle back
and forth between this and the main window, each is updated.
This window displays reports and error messages. The playing process
uses an sproc() call to create a child to play the sound. The process id
of this child is listed in the window. In case something gets munged and
you can't kill the child from the interface, you can always kill it from
a shell with the kill command, using the displayed number. (This normally
should not happen). This window will also report syntax errors from the
playnotes, although these can sometimes by cryptic.
It is not necessary to use this window, but it can provide some extra
gain control. It is opened by clicking on the tab at the bottom of the
main window. You can also turn tracks off and on by clicking on the track
number atop each set of sliders. The changes you make only go into effect
a few seconds after you make the change, due to the large size of the intermediate
sound buffers used to do the mix.
The track window is quite useful for balancing sounds, and for
preparing very complicated mixes which might exceed the throughput capacity
of the machine. You can turn selected tracks off and on, balance them against
each other etc.
The individual tracks gains are multiplied by the gains of the playnotes
in that track to effect the input amplitude.
The output track controls can be used to control the gain of the overall
mix.
If you save an rt script it will save the current state of the input
and output tracks, and the off/on state of the individual tracks.
Tracks can also be turned off and on from the main window by using the
buttons immediately below the information window.
The left mouse button will drag the slider of the channel of the track
it clicks on, and the middle mouse button will drag both sliders simultaneously
for that track, at the value of the track it selects.
Gains can be shown in decibels by clicking on the db box in the lower
left hand corner.
If you click on the little blue box below the DB box, you can control
tracks independently from the keyboard. Keys 1 through 0, - and =, will
increase tracks 1-12 respectively, and keys q,w,--> ] (just below) will
decrease the tracks. Keys g and b will increase and decrease the output
gain, respectively. In order to move several tracks simultaneously, you
have to repeatedly click on their keys. Holding down one key will repeat
that track but lock out the other keys. I don't know how to sove this,
and need an X-pert to show me how. When I can solve this problem and also
figure out how to make the response more immediate the next step will be
to record track motions. If you click elsewhere in the window you will
have to click the blue box again in order to reactivate this feature. s On the lower right hand side of the window there is an autoscroll switch.
Toggling this on will cause the window to jump to a location in which the
moving cursor is always visible.
The zoom in and zoom out buttons will give you differently
scaled displays. The default is zoom in. When you are zoomed out
the smallest visible duration may be about 1/4 inch wide. In the zoom
out case the time and duration of each note will not be displayed.
Holding down the <shift> key while using these buttons will zoom you
in and out by 50% (rather than 10%).
If you have a comment after a playnote: e.g.
The font for the text in the track map window can be set in the resource
file by adjusting the track_map_font parameter, listed towards the
top of the file. To get a complete list of available fonts use xlsfonts.
If you comment out that line in the resource file (the comment character
is !), the font will be iris10.
The shape of the symbols in the track map window reflects the amplitude
modifications you have made in the playnotes with the amp() command.
The color or the symbols reflects the gain as stated in the gain= command.
Bright tan is light and dark brown is soft, with various shades in between
reflecting gains in between. This color scale can be modified by editing
the resource file and changing the values for redlow and redhi, bluelow
and bluehi, and greenlow and greenhi.
You can then play the playnote or the soundfile alone, with the play
buttons to the left of each field. In the case of the playnote, what is
happening is that you are playing the mix with all the other tracks turned
off, and skip and end set to that playnote's time. Note that if you have
a small timescale, and a lot of tracks and sound, there may be a delay
before it starts to play. (Skips are much faster with larger timescales.).
The only way to stop either of these plays is with the stop button, or
command-L. If you hold down the shift key while clicking on the play
button, the note will start to play at the time specified in the skip field,
rather than at the beginning of the note. This way you can play selected
portions of a playnote as well.
There are several causes of potential confusion when using this page.
First, the reload playnotes button may not erase old playnotes
from memory, but they may not appear on the map. If, for example, you change
tracks, the playnote may still live on the old track as well but will not
appear on the map (unless it previously overlapped the end of an earlier
playnote, in which case it will not sound). The only way to make sure that
your map is accurate is to hit the reload everything button.
This is probably not necessary, however, if you are just changing amplitude
characteristics, or making the starting time a bit earlier.
If, on the main page, you loaded the playnotes after having made a selection
in the playnote window, what you will see is the result of that selection.
Changes in any playnotes will, of course, be copied back to the playnote
script, whether or not a selection is in effect.
Another potential source of confusion is in the case where you want
to simply reload one playnote in the main window by selecting that playnote
and hitting reload playnotes. In this case only that playnote will
appear in the track map, although all the other playnotes selected previously
will still sound. (This should be fixable, but my brain hurts right now.)
This window makes it very easy to fine tune characteristics of a mix.
You can turn tracks off and on with the buttons to the left of each track,
and start at any time in the mix simply by clicking the left mouse button
at that time, and then make changes in the playnotes and try again.
When a track is turned off, all displayed playnotes for that track will
appear partially greyed-out as a reminder.
While in the Track Map Window, if you ask to edit a soundfile (Alt-e,
or Edit Input File, from the Soundfile menu) you will get the soundfile
which is currently listed in the soundfile textfield on the lower right
of the window.
A potential source of confusion exists when you have used the offset
and/or the addoffset feature in the playnote window. Remember that
when you enter new playnotes the starting time will be incremented to whatever
the cumulative offset is at that point for the mix, and the displayed playnotes
will reflect this.
You can also move playnote by using the arrow keys, which will move
them up or down one track at a time or back and forth in time, in .1 second
increments. This feature merely requires that you have shift-clicked on
the playnote on the playnote you wish to move with the left mouse button.
When you are done, left-click anywhere in the window to redraw the playnote
in its new location. Again, you have to reload playnotes or
load
everything after you are done.
The duration of playnotes can also be adjusted with the mouse.
To do this, hold down the <shift> key and click on the playnote with
the right mouse button. Dragging to the right and left will
adjust the duration of the playnote. As you do this, the value for
dur=
changes in the playnote window below. Note that you cannot drag between
tracks using this command. As always, you have to reload playnotes
after you are done (load everything would never be needed for this operation).
The gain of playnotes may be adjusted up and down with the keypad '+'
and '-' keys. To do this, select a playnote using <shift>-leftmouse.
The '+' key will increase the gain by 1 dB; if the <shift> key
is down as well, the gain will increase by 3 dB. The '-' key will
reduce the gain in the same fashion. As you do this, you will see
the values for gain= change in the playnote window below.
Note that the relative colors of the various playnotes will also change
to reflect the new gain relationships.
1) Load Everything. Mashing
this button does the following: it opens and stats all the soundfiles listed,
it reads in the playnotes, it reads the timescale, the number of tracks,
and the state of the individual input and output tracks. It also allocates
memory. When it is yellow it indicates that no relevant information is
in memory, that memory is not allocated, and that nothing will happen.
Since rt uses a lot of virtual memory, miniaturizing the app in
this unloaded state makes it possible to have a very large number of these
apps running simultaneously. (See the note on virtual
swap space).
2) Kill everything. Mashing
this guy will simply kill the current driver and stop everything that is
going on. It resets the app to a zero state, and releases virtual memory.
It is useful to hit this when things get confusing or out of hand.
3) Reload Playnotes. This
button assumes that the driver is already running, and it will simply read
in either the entire contents of the Playnote Window, or a selection
you have made with the mouse. This is useful if you want to add a note
or alter a note. You have to be careful about allocation here, however,
since a track can only play one note at a time you may wipe out something
that was already there. (This is often what you want to do.) When things
get too complicated it is usually best just to restart the driver.
4) Play. This will play the currently
loaded mix, using the times given in the play from, play to forms.
The button is inactive unless the 'load everything' button has been hit.
5) Stop. This will stop the currently
playing mix, but will not kill the driver. Sometimes it takes a moment
or two to kill the playing process. This button is inactive unless
a mix playback is in progress (or is paused).
6) Pause. This will pause the
currently playing mix. Currently it may take a few seconds to pause.
This button is inactive unless a mix playback is in progress.
1) Timescale. This describes the
quantization of the mix. The driver works by quantizing time into ticks
of this length. At each tick it stores information about who is playing
where, what, how loud, etc. Each tick thus contains a complete record of
the state of the mix for that moment. The data in the Playnote window
is
used to set this information up. The default timescale is 1/100
of a second. This means that sounds, amplitude alterations, etc, can alter
states with this granularity. In other words, if you want a note to enter
at time 1.005 and your timescale is .01, your sound will enter at
time 1.00. The sounds will play to whatever duration you specify, regardless
of the timescale, and transitions over that time will be smooth,
no matter what the timescale is. The maximum timescale for a 44.1k
mix is about .8 seconds. The limitation is caused by the size of the buffers
being sent to the DACs. .01 is a good timescale for most purposes.
Larger timescales will result in somewhat more efficient mixes since
there is less reshuffling of the disk and data. (Changes in amplitude and
frequency are done on a sample-by-sample basis, however, regardless of
the timescale, so larger timescales should not normally result in clicks.).
There are certain conditions, however, in which a smaller timescale may
result in better performance. In particular, if a great many files are
being mixed, a smaller timescale seems to allow disk overhead to be distributed
more evenly over a period of time. With a large timescale, at each grain,
a lot of disking has to be done and this might cause interruptions. There
are currently a maximum of 200,000 ticks allocated, so with a timescale
of .01 you are limited to a play of 200.000 * .01 = 2000 seconds. Increasing
the timescale increases the amount of time available. It is often worth
experimenting with the timescale. If you have a mix which only contains
a few large files, a larger timescale will be more efficient. Skips into
a file are much quicker with a larger timescale. 2) Tracks. This is the maximum
number of tracks you will be using. The default is 8 but up to 12 tracks
can be used. If you are using less than 12 (the absolute maximum) you may
want to specify fewer to gain slightly greater efficiency, but I don't
think that this normally would make much difference. See the manual page
on the Tracks Window 3&4) Play from.., play to. Very
often you will only want to hear selected portions of a mix. You may enter
the starting and ending times in these two forms. You don't need to reload
or restart the mix to do this, just change the times and hit play. If you
put a very high number in the play to form, it will just play to
the end of the mix.
5) Timer. Displays the current time
of the mix in 1/10 second increments. If you skip in the timer will skip
in as well. If you turn soundfiles off while playing, however, the timer
if there are periods of time in which no soundfile is active, since this
time will not be updated . The timer also appears in the trackwindow 6) Peak Amplitude. Displays
the current peak amplitude of the buffers just before they are sent to
the DACs. The fields below the peaks show the maximum so far, and can be
reset by the reset button.
If you make any changes in the names, conditions, or numbers of soundfiles,
or the timescale or maximum number of tracks, you will have to restart
the driver (kill everything, load everything). Most other changes, such
as altering the times of a mix in the control window, some changes
in the playnote window, turning soundfiles off and on in the soundfile
window, making any changes in the tracks window, can be done
without restarting the driver.
The entire state of a mix can be saved to a simple ascii file with file
type .rt. This is done from the File menu, and if it is a
new mix, you will be prompted for a filename. Similarly, you can restore
a mix by first opening rt with the rt script specified in the command line,
or by opening rt and then loading in the script from the files menu.
When you finally want to save a mix, before hitting the play button
go
to the Soundfile menu and select the Write NeXT soundfile
or the Write WAVE soundfile option. It will prompt you for a soundfile
name. Now hit the play button. What will be written
to disk is exactly what you would have heard had you hit the play button
alone.
As the mix proceeds, you will be updated on its progress via a dialog panel,
and you have the option of interrupting the mix at any point. The
nux-to-disk option resets after each play command, so you will have to
repeat the above steps to rewrite the mix to disk.
It's up to you to make sure you have enough disk space. (hint: learned
the hard way--if you fill up your disk don't try to save your rt
script,
it will just write an empty file.)
To edit a soundfile place the cursor anwhere in the name of that soundfile,
in the soundfile window, and go to the edit input soundfile selection
in the Soundfile menu. The sound editor program specified by the
Rt*soundEditor
resource will be invoked on the soundfile, if the program is installed
on your machine. Otherwise you will be alerted that it can't find
it. Under Linux, the current default is Douglas Scott's "mxv"
program.
You can also edit a playnote, as modified by its parameters. To do this
you need to select a playnote in the Track Map Window with the middle mouse
button. The playnote will then appear in the form at the bottom left of
the Track Map Window. Then, Alt-K will write that playnote's sound
in /tmp, and the sound editor program will open that file. When you are
done you will be prompted to see whether or not you want to save the soundfiles
in /tmp. Note that any changes made to the playnote soundfile with the
soundeditor will have no effect on the resulting mix.
Rt will now read wave, aiff and NeXT/Sun soundfiles The last column
in the soundfile window displays the format after the file has been loaded.
"w" for wave, "a" for aiff, and "n" for NeXT/Sun. In the Soundfiles
menu there are options to alter the headers. To select a file for alteration
place the cursor anywhere in the name of that soundfile, and call up the
menu selection. You will not be allowed to write an aiff header if it would
destroy any sound data. . (This will only happen in the case in which an
aiff header is requested for a NeXT soundfile where the header is the minimum
size, 28 bytes). The new file format will be shown next time the soundfiles
are loaded with the "load everything" button.
Rt works by dividing time into ticks, as specified in the timescale
window. Each tick contains information about what is going on at that track,
at that time. For each track there are 200000 ticks allocated for each
of 10 parameters. Normally only a small number of these locations are used,
but when the memory is allocated the operating system checks to see if
that much virtual memory is available. So, for 12 tracks, you would have
to have 12*10*4(sizeof(int))*200000 = 96 megabytes of virtual memory. In
the (unlikely) event that you had all 12 tracks going for all 200000 ticks
and were setting all parameters all the time you *might* start to use up
your swapfile. (I've never seen this happen, but it's possible). The following
are recommendations regarding use: First, when the 'load everything' button
is in its yellow state, no virtual memory is allocated. This means that
you can have lots of different rt jobs running at the same time, as long
as the sum of the requested virtual memory doesn't exceed what you have
available. Second, use only as many tracks as you need, and specify the
limit in the tracks window. This way less virtual memory will be requested.
Finally, if you are going to 'minimize' an rt job, be sure to 'kill everything'
before you do. This way it will take up only a few meg of available memory.
On a 200 mhz INDY, running 5.3, reading from a Seagate Barracuda drive,
the program seems to be able to play 12 44k stereo files at one time with
a timescale of .02. The larger the timescale, the fewer function calls
there are. Reading files over NFS is generally going to be much slower.
Look ahead buffers. There are two internal
buffers which are used. One is in the driver, and the other is in the DACs.
The driver buffer is fixed at 64k, but the size of the DAC buffers can
be adjusted to be between one and six 32k buffers by using the slider on
the track window. The effect of decreasing the number of these buffers
is that changes made by moving the sliders for the individual tracks and
the output tracks will take effect more quickly, as will the stop and pause
buttons. The negative effect is that there may be glitches since the DACs
may run out of data. For complicated mixes it is best to use a larger number
of buffers. You don't need to restart the driver to reset the buffers,
but they will not take effect during a play.
You have three soundfiles you want to work with. Let's call them talk,
whistle and knock.
1) Enter them, in this order, in the soundfiles window.
talk is a long soundfile and you want to use selected parts of
it. whistle is the sound of someone whistling which you want to
intersperse here and there, and knock is a simple annoying knocking
sound.
2) you might have the following list in the playnote window
playnote(snd=1,track=1,amp(0,1,4,1,5,0), end=10) //give me 10
seconds of talk fading out at the end
playnote(snd=2,track=2,amp(0,0,1,1,2,0), skip=2,end=5,transp=-1,at=5)
// give me times 2 through 5 of the whistle, down 1 semitone starting
at time5
playnote(snd=3,track=3,at=3,pan=.1)
playnote(snd=3,track=4,at=3.2,pan=.6)
playnote(snd=3,track=3,at=3.4)// several knocks, assuming that
the knock is less than .4 seconds long
3) then hit, load everything, play and listen. You notice
that the balance and envelopes are all wrong so you open up the tracks
window and adjust the gains individually, listening to the sounds individually.
4) You decide that the transposition level for the whistling is all
wrong so you change the transp=-1 to transp=-2. Then select this line and
hit the reload playnotes button. Hit play and listen
again
5) You then decide that the whistling comes in too soon, so you adjust
its at=5 to say at=6, select, load and play. Now you notice
however, that the note still starts at time 5 and then starts again at
time 6. In otherwords you haven't wiped out the first reference to it.
The only thing to do now is to restart the driver by hittile the load
everything button..
6) You want to listen to the section from time 3 on now, so you type
3 in the play from form in the control window.
7) You decide it is good enough for your recording company so you write
the mix to disk and go to bed happy.
Many of the windows and buttons have SGI-style help panels. To activate
them put the cursor over the window and click the F1 key. This is probably
most useful for the playnote window, where the panel will show the complete
syntax for playnote commands. 1) As of this writing (1/5/97) I think that the rev function is broken.
2) If you turn soundfiles off during a play, and there are moments when
no soundfiles are active, the timer will not be updated properly. (This
may be a feature)
3) To turn a soundfile back on you have to move the cursor to a different
position in the word 'off'. (Due to the inadequacy of X callbacks).
4) Often you will see artifacts left over in the track map window after
moving playnotes, panning, etc. Just hit reload playnotes
and these will all go away.
If you know what you are doing you can change fonts, colors etc, by
editing the Rt resource file. If you don't like my color schemes, learn
how to use cedit and you can then specify any of 256^3 colors. If
you want to use your own resource file you should set the XUSERFILESEARCHPATH
environmental variable to %S%N and copy the Rt resource file to your current
directory. Thanks to David Jaffe for important additions to the NeXT port, which
will also make it into this port, eventually. Peter Velikonja did the major
work in de-NeXTifying the sound driver so that it would work without the
SoundKit. To Doug Scott for simplifying the use of the SGI Audio Library
for me. Thanks to Bill Schottstaedt for helping to grok aiff files. And,
most of all to Kent Dickey, whose class project in Cos/Mus 325, the driver
for this app, has helped out so many people since that time. (I should
have given you an A+)
Paul Lansky 1/5/97
Revised by Douglas Scott 06/2002
After every 25 characters are altered or entered in the playnote window,
a complete backup of the current rt script will be saved in the /tmp directory.
It will be called <filename>.bk if you opened a script called <filename.,
and untitledxxxx.rt.bk, where xxxx is some arbitrary number, if you a creating
a new unnamed script.
Soundfile Window
The gain change will take place during a play with delay subject to
buffer size -- see lookahead buffers
-- ssif you change it with the slider. Otherwise it will only take effect
on the next play.
Playnote Window
Some Special Features
Use this feature to allocated clumps of playnotes to specific groups
of tracks when you are not specifying the track number in playnotes, but
rather using the automatic feature. This will improve the track placement
hygiene.
Playnote Window -- Large View
Information Window
Tracks Window
An Experimental track feature
Track Map Window
Display in the Track Map
Window
Once playnotes have been loaded with the load everything button,
or with the reload playnotes button, this window will display
a timeline layout of the tracks and soundfile segments on each track. Below
each soundfile are three numbers indicating starting time, duration, and
soundfile number for that segment. While playing, a timeline will march
along, and the time will be displayed in the lower left hand text field.
If you click the left mouse button anywhere in the window, the next play
will begin at the time you clicked and the starting time will be displayed
in the right of the two text fields. (You can also simply type a number
in that window to force a starting time.) Sometimes you will have to wiggle
the mouse to get the timeline and the listings to start up. Remember that
if you are skipping into a mix, there may be a pause before it starts to
play. If you click the left mouse button while a sound is playing it will
stop playing. While playing if you click the right mouse button anywhere
in the window, the mix will start to play at this point. If it is already
playing it will stop and start again at the point at which you clicked.
This enables you to jump around a mix. Again, there may be some slight
delay, depending on your timescale, and the size of the timeskip.
playnote(track=1,snd=2,at=23,skip=23)//big noise
or
playnote(track=1,snd=2,at=23,skip=23)big noise
this comment will appear above the rectangle representing the track.
If there is no comment, the soundfile name will appear, stripped of its
full path. Sometimes, depending on the length of your comments, track spacing
and filenames, the text of one playnote may run on over the text of another.
There are also cases when the end of one playnote will overlap the beginning
of another. These should be obvious. (I'm going to leave it this way for
a while until we get more experience with this feature.)s
Editing existing playnotes
in the Track Map Window
Attributes of a playnote can be changed from this window. If you
click the middle mouse button on any playnote bar, the playnote script
for that note will appear in the textfield below and the soundfile name
and number in the lower field to the right. You can then edit it. Do not
hit <return> after editing the field -- doing this will insert a new
playnote in the list with the current values. When you press the reload
playnotesor reload everything button the changes
will take effect. If you want to make a bunch of changes in several playnotes
you can enter the changes into memory by just editing them all, then hitting
either of the two reload buttons to make all the changes at once.
Visual editing in the trackmap window
Adding new playnotes
from the Track Map Window
You can add new playnotes to the mix simply by typing them in the
field, hitting <return>, and, after all the playnotes you want to add
have been entered, hitting reload playnotes, or reload everything.
The
default program state assumes that what you are entering in this field
is a new playnote, unless you have clicked the middle mouse button on an
existing playnote bar. This sets the state to modify. The state
will be reset to the default add when you hit reload playnotes
or reload everything.
Visual editing of playnotes in the
track map window.
Playnotes can be moved around with the mouse or the arrow keys.
To move a playnote with the mouse, hold down the <shift> key and click
on the playnote you want to move with the left mouse button. Note that
the text and end-bars for this note are now displayed in red. This
indicates that this is the playnote selected for editing. Now drag
the playnote with the mouse. You will see a white version of the soundfile
move with the mouse. You can change both its start time and track number
this way. When you release the mouse button, the original image of
the soundfile will disappear and the new one will assume its color. You
can move around any number of soundfiles before hitting the reload playnotes
or load everything button to put the changes into effect. If you
move soundfiles to the left, later in time, you should use the load
everything button. As you move the soundfile you will see the values
for at= and track= change in the playnote window below.
Controlling track playback from
the track map window.
Tracks can be turned on and off from the track map window by first
clicking with the left mouse button in the window to gain focus, and then
using a combination of the middle mouse button and the letter z to turn
tracks off, and the middle mouse button and the letter x to turn tracks
on. You can drag the mouse up and down to select groups of tracks. When
a track is turned off, the line under that track will turn white, and all
playnotes on that track will appear partially grayed-out. You can turn
all tracks off with Alt+u and turn all tracks on with Alt+t.
Various Buttons
Information Fields