This is a snapshot of CSOUND Version 2, currently running at the MIT MediaLab on DEC-RISC machines (under Ultrix 4.2), on SGI Indigo's (under 4.1 OS), on SUN's (under 4.1 OS), and on Macintosh's (under THINK_C's Lightspeed C 4.0). It is also known to run on NeXTs, and probably on many others. If you plan to run on a Unix system, then these sources are what you need. If you plan to run on the Mac, you should ftp the .hqx versions of same. CSOUND has at various times also run on HP Bobcat workstations, but the variants required have not recently been checked out. Maybe someday soon. All system differences are controlled by the top-level Makefile. By editing this Makefile you can include X11 windows software, define the soundfile system you are addressing, and include certain device-dependent I/O drivers. The Makefile can also control the speed of audio processing: electing single precision floats will double your speed with no loss in fidelity; even greater improvements will result from any hardware accelerators present. Check your hardware and software options carefully while editing this Makefile. The top Makefile requires you to specify the libraries which will receive your executables on successful installation. Once Makefile is suitably modified, installation requires just five commands from within the top directory: make all make install make manpp make clean If you experience difficulty, the first 'make all' can be done in separate stages: make csound ;the top-level csound compilation make anals ;subdir of analysis programs (adsyn, lpc, pvoc) make utils1 ;subdir of standalone score decoding, sort & extract make utils2 ;subdir of file managing programs (sndinfo, scale) If you get only as far as 'make csound' you can immediately start using Csound since you have a COMPLETE EXECUTABLE. This includes a -P (preprocess) option for generating the sound analysis control files suitable for ADSYN, LPC and PVOC audio reconstruction (see the new Manual). It likewise already contains the SCOT, SCSORT, and EXTRACT score manipulating programs. The subdirectory phases 'anals' and 'utils1' create stand-alone versions of the analysis and score manipulating programs. CSCORE, and the utils2 file managing programs, are available only at this standalone level. Problematic routines can be bypassed by commenting out the offending lines in the lower Makef files. NO OTHER FILES SHOULD BE MODIFIED AT INSTALLATION. Finally, use 'make clean' to remove old intermediate files & conserve storage. Sound is written to a file in a predefined Sound File Directory, declared by setting an environment variable SFDIR (eg. 'setenv SFDIR /usr/sound/logname'). You can read/write any of 6 Soundfile formats: 8-bit char, 8-bit a-law, 8-bit u-law, 16-bit short ints, 32-bit long ints, or 32-bit floats; two of these (16-bit int, 32-bit int) are also readable/writeable as AIFF files. If you install with SFIRCAM defined, soundfiles generally have a 1024-byte header. Installing with SFSUN41 defined will enable use of SUN 4.1 multi-media conventions. Since sounddisk drivers, RECORD and PLAY commands are generally hardware dependent, this package recognizes only some of these. If you have a VAX and Digital Sound Corp DAC's, you should import the IRCAM sound-file system and their play command. If you have a DECstation and the LoFi board, you can should install with -DDEC. If you have an SGI machine with audio board, you can should install with -DSGI. A SUN installation can read/write u-law soundfiles by invoking /dev/audio in the command line, or can write other formats using the 4.1 library. I have made a small header optional on all control files for LPC and PVOC, so that srate, npoles, etc. can be automatic (see the lpc.h file). Csound will read the header, then step over it no matter how long it is. The idea is to document your lp files by expanding the text in the header. See the 'man' pages on this topic. The directory 'analfiles' contains sample files for adsyn, lp and pvoc. See the README in that directory, and how to set the environment variable SADIR. The directories 'scorefiles' and 'morefiles' contain orchestras/scores of musical examples, included for your initial assurance. Some of these are transcriptions of traditional pieces, likely to run in realtime if your machine is fast enough. Others are original scores by expert users (James Dashow, Jon Nelson, Richard Boulanger); usual copyrights apply. The directory 'midifiles' contains the beginning of a MIDI performance direction, including a Beethoven Pastoral transcription; see the README in that directory. The directory 'testfiles' contains smaller orchestras/scores of tests I've made in passing. You might want to check these against the manual for syntax and form; the task of creating bigger orchestras and better scores is left to you. The Csound Users' Manual is now maintained only as a Microsoft Word document. It is available over ftp as either a Word or PostScript file. Download either. The manual is very detailed (it's a Reference tool, not a Primer), with the goal of giving an accurate account of each facet of the Score and Orchestra languages. This new version, however, does include some tuturials and examples as Appendices. I am maintaining a BULLETIN BOARD MAILING LIST for Csound users. This is an entry in our /usr/lib/aliases that I normally use for Csound announcements, and which I'm making available for user intercommunication. Since this has induced some confusion in the past, it is now RENAMED: csound-BB@media-lab.media.mit.edu BE AWARE this BROADCASTS to everyone in the mailing list. Use network etiquette! For ADMINISTRATIVE REQUESTS (have your name added-to/removed-from the above, order hardcopy manuals, etc) mail to: csound-request@media-lab.media.mit.edu Report bugs to: csound@media-lab.media.mit.edu Barry Vercoe August, 1992
