This is, by the way, NOT the same as the "well-tempered" scale which is used by J. S. Bach, and is the title of his famous collection. In ancient music, a system is "well-tempered", if you can play all major and minor scales in it (without dying of ear pain). So there are MANY "well-tempered" scales (Kirnberger, Werckmeister and others) in this time and also before. NONE of it was "equal-tempered". In an equal-tempered scale, one major scale sounds like any other, and so do the minor scales. The only difference is the absolute pitch. So there are many "well-tempered" tunings in history, but there is just one "equal-tempered".
Schoenberg, Arnold. "Method of Composing with Twelve Tones Which are Related Only with One Another" ("Komposition mit zwölf nur aufeinander bezogenen Tönen"). See his lecture "Composition with Twelve Tones"(1941). Style and Idea. ed. L. Stein. London: Faber and Faber Ltd.,1975. p. 218.
 I did this in TclCsound, too. It can be downloaded ("Stimmungen") from www.joachimheintz.de/software.
 You just must not check the "Ignore CsOptions" checkbox in the Configurations Dialog. You should do something like this:
 In QuteCsound you actually do not need this. You can choose your output device in the configuration dialog, or even browse to the one you would like to choose. See the screenshot above.
 If you just want one keyboard to receive Csound, you can choose your device using -M0, -M1 etc (also here QuteCsound allows you to browse your MIDI devices in the configuration dialog).
 You can also try lower values for -b and -B (e.g. -b64 -B256 or -b32 -B128 or even -b16 -B64). There is an intricate interaction of the ksmps (in the orchestra header), the -b (software buffer size) and the -B (hardware buffer size) flag. It is well described in the relevant section of the manual ("Optimizing Audio I/O Latency"). Refer also to the explanation by Victor Lazzarini regarding the "full duplex audio" mode when using realtime audio in (-iadc and -odac) in http://www.nabble.com/pvsbufread%2C-ksmps-and-ghosts-over-p3-to23684045.html#a23707140.
 The exact value is 1.9549576.... You can calculate this value by starting from one pitch (say 200 Hz) and then comparing the pure fifth (which has by the ratio 3:2 a frequency of 300 Hz) with the equal-tempered fifth (which is calculated as 200 * 2^7/12 = 299.6614... Hz). Then you get the cent difference of both frequencies by the formula 1200 * log 2 (freq1/freq2). In Lisp: (* 1200 (log (/ 300 (* 200 (expt 2 7/12))) 2)).
 The absolute cent values are: 0 - 76 - 193 - 310 - 386 - 503 - 579 - 697 - 773 - 890 - 1007 - 1083. See Klaus Lang, "Auf Wohlklangswellen durch der Töne Meer, Temperaturen und Stimmungen zwischen dem 11. und 19. Jahrhundert", Graz 1999, p. 68, now online available under http://iem.at/projekte/publications/bem/bem10/.
 The negative number for the Gen-Routine (-2) avoids the normalization of the values we put in the list. The negative number for the size of the function table (-12) allows Csound accept non-power-of-two tables.
 Without the widget specifications. They are included in the attached example files.
 A better one would probably be to calculate the frequency of a pitch without comparing it to the cpsmidinn values. This is more or less the way we will go in tune_07.csd, where we have more than 12 steps per octave.
 The absolute values are: 0 - 114 - 204 - 294 - 408 - 498 - 612 - 702 - 816 - 906 - 996 - 1110. Klaus Lang (see above), p. 41.
 Sambamoorthy, P. South Indian Music. Book IV, Second Edition, Madras 1954, pp 95-97.
 Probably there could be a more elegant solution, but this seems to do what it should do.