hetro
takes as input a file containing amplitude samples of some
sound over time (it is assumed that the samples are evenly spaced
in time) and decomposes that sound into a set of harmonically
related sine waves with time varying amplitude and phase.
Below are listed the flags and their meanings. All flags must be followed by a value with no intervening space between the flag and its value.
The following is an example call to hetro:
This will analyze 2.5 seconds worth of samples in infile, which was created with a sample rate of 25kHz, beginning 0.5 seconds into the sample file. Hetro will analyze the first 10 harmonics of the sound, and output 256 points for each of the time-varying amplitude and frequency pairs into the file adsyn.4 with a maximum amplitude of 3000. The LPF is disabled.
Due to the large amount of calculation required to do the analysis, hetro runs fairly slowly. Think carefully about how many harmonics and how long a sample file is necessary to analyze.
Tom Sullivan, a.k.a. sullivan@ems.media.mit.edu.
The adsyn unit generator in Csound provides a means of synthesizing complex, time-varying timbres using the technique of additive synthesis. adsyn works from a control file called adsyn.m , where m is an integer (see Csound manual). The file adsyn.m specifies the amplitude envelope and frequency path of each partial that makes up an additive complex tone. The information is in the form of breakpoints (time, value, time, value, .....) using 16-bit integers. Time is given in milliseconds, amplitude values are given in the range of [0, +32767], and the frequency is given in cps (Hertz). Breakpoint data is exclusively non-negative; the value -1 in the control file marks the beginning of an amplitude envelope and a -2 in the control file marks the beginning of a frequency set. A time value of 32767 followed by a replication of the last desired amplitude or frequency value is always included at the end of each envelope or set.
A partial is defined by a pair of sets: an envelope set and a frequency set. The sets may appear in any order but there must be an equal number of amplitude and frequency sets in the file. For example, you may alternate amplitude and frequency sets, list all amplitude sets first then all frequency sets, or do a few of one kind then a few of the other, but in the end there must be the same total amount of each.
There can be any number of breakpoints in each set but adsyn associates the first found amplitude set with the first found frequency set, second with second, and so on.
The format of the adsyn.m is as follows:
DATA EXPLANATION
___________________________________________________________________________
-1 time1 value1 ......timeN valueN 32767 valueN ; amplitude partial 1
; breakpoints
-2 time1 value1 ......timeM valueM 32767 valueM ; frequency partial 1
; breakpoints
;in any ordering of
-1 time1 value1 ...... ;amplitude and frequency
-2 time1 value1 ...... ;but ordering of amplitude
-2 time1 value1 ...... ;partials must be the same
-1 time1 value1 ...... ;as ordering of frequency
;partials.
The Hetro program outputs to an adsyn.m file in this format but alternates amplitude and frequency sets in increasing partial order.
Hetro
is based on Sullivan's `hetro' program (described elsewhere), rewritten to
operate specifically with NeXT soundfiles. The original hetro
expected 16-bit integer sound data, whereas Hetro will work with any valid
NeXT sound file format. Hetro may be invoked from the workspace, a shell
window, or another application. It consists of several form cells
for parameter entry, two buttons and a status display cell. Before
any analysis can take place, a valid soundfile (.snd) must be opened. This
is accomplished either by clicking on the `File' menu entry or by typing
<command>-o. Also, an estimated fundamental frequency (> 0) must be entered
in the form so marked. If these conditions are met and Hetro agrees to
analyze the soundfile (Hetro does not analyze multi-channel files, for instance),
the analysis will proceed according to the entered parameters. The results
of the analysis will be recorded in a file identical in path to the soundfile,
except with a `.hetro' suffix. If write permission is denied in the
directory containing the soundfile, then the soundfile will have to be copied
or moved before being analyzed. If the `duration' and `filter' cells contain zeros
or non-numeric data, their defaults (described elsewhere) are taken. The
`analyze' button initiates the analysis; the `sound info' button puts up a
panel displaying information on the currently opened soundfile, such as
data format, size and sampling rate. The `help' item in the menu displays
the hetro man pages on a scrolling text display.
Pete Yadlowsky, UVa, pmy@virginia.edu