siggen.conf - the siggen configuration files
As from siggen version 2.3 onwards a versatile configuration file scheme has
been introduced. It allows parameters for the siggen programs to be specified
either across the board, or specifically for particular programs.
Three possible configuration files can be used: a LOCAL config file (usually in
current directory), a HOME config file in user's $HOME directory and a GLOBAL
All the programs are compiled with the names of the config files built in. The
filenames are set in the config.h header file and can be changed. The LOCAL
and GLOBAL config files are specified by the settings of:
- #define DEF_CONF_FILENAME ".siggen.conf"
- #define DEF_GLOBAL_CONF_FILE
And can be set to any file name or to NULL to disable the file. The HOME config
filename is created using the $HOME environment variable and the
DEF_CONF_FILENAME together, i.e. using the above, the HOME config file for a
user whose home directory is at /home/jj, would be
The config files do not have to exist. If they exist and are readable by the
program they are used, otherwise they are simply ignored.
The config files are always searched for configuration values in the order
LOCAL, HOME, GLOBAL. This allows a scheme where the sysadmin sets up default
config values in the GLOBAL config file, but allows a user to set some or all
different values in their own HOME config file, and to set yet more specific
values when run from a particular directory.
If no configuration files exist, the programs themselves provide builtin default
values (see config.h etc), and most of these values can be set by appropriate
command line switches and flags.
A configuration value has a name and a value, and values for all programs are
set by simply entering a line in the appropriate config file where the first
word is the name, followed by arbitrary spaces/tabs, followed by the value.
The value is all the rest of that line. e.g. to set the global default
samplerate of 44100 samples per sec, the following line would be entered in
the GLOBAL config file:
- SAMPLERATE 44100
Config value names are case insensitive.
A config value can be set for a specific program, by prefixing the config value
name with the program name and a ':'. e.g. to specify a samplerate of only
8000 samples per sec for the tones program enter
- TONES:SAMPLERATE 8000
in the relevant config file. If both lines above were in the config file, all
programs except tones would use a samplerate of 44100, and tones would use
You do not have to specify all configuration values in the config files. If a
particular value is missing, the programs will simply use their builtin
defaults (see config.h etc).
Configuration values set by command line switches or flags take precedence over
values in any of the config files.
- the programs do not have their 'name' built-in, but use the
name they were invoked by. So if you change the name of a program,
remember to change the config file entries. However this does means that
by using links to a program, it can be made to pick up a different set of
configuration values, depending on the name it is invoked by.
A sample config file is provided in ".siggen.conf" in the
distribution. This may also be at /etc/siggen.conf . Any line whose first
non-whitespace character is a '#', is a comment line and is ignored.
Not all of the siggen programs use all the values described here. See the
relevant man page for which values are used by which programs.
- In all programs except tones and fsynth, channels specifies
the number of output channels to use, i.e. 1 for Mono and 2 for
- For tones, channels specifies the number of 'voices' on
which tones can generate different waveforms before mixing them into the
one output channel.
- For fsynth, channels specifies the numbers of separately
configurable oscillators used to mix the single output channel.
- The Digital to Analogue Converter (or PCM or DSP) device on
which to output the generated samples. This must be a real OSS PCM device,
otherwise the ioctls used will fail.
- The number of Audio Buffer fragments to configure in the
driver. The interactive programs respond to changes made to parameters
from the keyboard immediately, but data will be buffered in the driver in
the buffer fragments. If the amount of data buffered is too much then
there will very noticeable delays before the output sound is altered.
Against that, insufficient buffering may mean that there is not enough
data buffered for output to cover the time when other processes are being
run by the scheduler. The programs set the buffer size to the nearest
power of 2 to give aprox. 100millisecs of sound. Hence if FRAGMENTS is set
to 3, there will be aprox. 0.3 secs worth of sound buffered for output. On
a lightly loaded fast machine this, or 2, should be sufficient. To cover
periods of heavy load or on a less powerful machine use 4 or 5. But
remember the interactive programs will appear sluggish in responding to
- The number of samples per second to use. If output is to
the DAC then the DAC device is set to output samples at this rate.
BEWARE: not all cards can support all samplerates. SoundBlasters are
fairly flexible in this respect. Other cheaper cards are not. Indeed some
cards can only handle a very restricted set of related samplerates e.g.
11025, 22050, 44100 & 8000, 16000, 32000, 48000. When writing to
DACFILE all programs will attempt to set the samplerate given, but use the
actual samplerate the device used. Use the verbose command line flag to
check actual samplerates used.
Some common samplerates used are:
- is the samplerate used in the phone system with 8 bit
samples, and is adequate for voice range frequencies.
- is the samplerate used in audio CDs
- is the samplerate used in DAT systems, I think, and for
much professional kit.
- is also used, but I forget where, minidisc?.
- In general, the higher the samplerate the larger the memory
and processing requirement, but the higher the frequency range and the
more accurate the signals generated.
- Number of bits per sample. Only two values are allowed
currently, 8 or 16.
- bit samples are unsigned, with decimal value 128 being the
- bit samples are signed little endian values, i.e. the least
significant byte comes before the most significant byte either in a file,
or in the byte stream to an output device.
- If samplesize if left completely unspecified, then all
programs will attempt to do 16 bit samples to DACFILE, and if that isn't
possible will do 8 bit samples. Or if writing to a file, 16 bit samples
will be written.
- sets verbosity level.
- is quiet
- is be a bit verbose (equiv. to -v switch)
- is be very verbose (equiv. to -vv switch)
- if set to a non-zero value, then the VI cursor moving keys
"HJKL" are enabled.
Copyright 1995-2008 Jim Jackson
- sgen(1), swgen(1)
The software described by this manual is covered by the GNU General Public
License, Version 2, June 1991, issued by :
- Free Software Foundation, Inc.,
675 Mass Ave,
Cambridge, MA 02139, USA
Permission is granted to make and distribute verbatim copies of this manual
provided the copyright notice and this permission notice are preserved on all
Permission is granted to copy and distribute modified versions of this manual
under the conditions for verbatim copying, provided that the entire resulting
derived work is distributed under the terms of a permission notice identical
to this one.
Permission is granted to copy and distribute translations of this manual into
another language, under the above conditions for modified versions, except
that this permission notice may be included in translation instead of in the