MIRAGEFREQ Sample Tone Generator Info: The MIRAGEFREQ program does essentially one thing - it generates a tone that matches the current sampling rate of an Ensoniq Mirage. It does this by requesting the current sample rate via MIDI (MASOS 2.0 must be booted up), and then calculating the sample rate and producing a tone on the Commodore SID chip. The reason this is important has to do with the way the Mirage handles short loops. To work correctly, the length of a short loop should be a power of two: one, two, four, etc. and the number of the first page of the loop should be divisible by the length of the loop. Refer to the Advanced Sampler's Guide for more details (especially Part IV). Since the Mirage can only sample at certain frequencies, we are required to match the sound being sampled to those frequencies, if we want to have an integral number of waves in a loop (we do). That's what MIRAGEFREQ does. Running MIRAGEFREQ is very straight forward. Simply run the program - it should indicate what it's looking for. Specifically, you need to have a Mirage running MASOS 2.0 and both MIDI in and MIDI out should be connected to your MIDI interface, preferably directly. (Connect out to in and in to out.) The program automatically detects a Passport Designs or Sequential Circuits (SCI) style of MIDI interface and handles the differences for you. MIRAGEFREQ will run on either a C-64 or a C-128 and on the C-128 in either 40 or 80 column mode (the same is true for this program). If there is any problem MIRAGEFREQ will continue to poll the Mirage until there is a response so you can simply run the program and do the cabling and booting thing as you get to it. Once the connection is made, the program will clear the screen and carry on. What MIRAGEFREQ does then is to display information that should be helpful when sampling. This includes the sample rate parameter [73], the actual sample frequency, the cooresponding note frequency as well as the name of the equal tempered note (A=440) that is closest. The note is shown with a +/- error percentage - this is actually in 'cents' which is 1/100 of a semi-tone. This means that -50% is right between the displayed note and the note just below it on the scale. This should prove handy in selecting values for the sample rate. You may also note that only two sample rates are very close at all to standard pitches: 69/Bb and 71/A. With this program, you can adjust your source or the sample rate for very good short loops. Also displayed are the Nyquist frequency and the input sampling filter frequency. This is to help avoid the dreaded aliasing. Aliasing is caused when something is sampled in time at too low a rate. Nyquist showed that to prevent aliasing, the sample rate must be AT LEAST twice the highest frequency component. The classic example of aliasing is wagon wheels in a western movie. As the wheels speed up, the movie frames (samples) appear to slow down, stop, and then to go backwards; this is aliasing. A similar result occurs when frequencies are sampled at or below the Nyquist limit (twice the highest frequency). As the sampled frequency approaches the limit, it will, when played back, sound a bit lower in frequency than expected. If the highest frequency is well beyond the Nyquist rate, it's harmonic relationship to the rest of the sample will be destroyed. This will often sound pretty nasty (though it can on rare occasions prove useful). There are two basic ways to prevent aliasing. First is to make sure that the sample rate is high enough - often difficult on a Mirage due to the limits of the hardware, and the second is to filter out most or all of the frequency components of the sound that are above the Nyquist rate. This is done on the Mirage with the input filter. The Mirage has a built in filter, as well as an option for an external filter (the optional Input Sampling Filter - ISF). Both are supported by this program, but since I don't have an ISF, I can't test that part of the program fully. The ISF also offers addtional, higher sampling rates. To the best of my knowledge, Ensoniq no longer offers the ISF so if you don't have one already, you'll have to dig up a used one. Filters are not ideal and setting one to the Nyquist rate will not be sufficient to prevent aliasing. In general, the frequency given is the 3dB point - the point at which the filter lowers the incoming signal by three decibels or essentially cuts the signal in half. Where to set the filter to be sure to eliminate aliasing is beyond the scope of this tutorial; your best bet is to experiment, using the numbers that MIRAGEFREQ provides. Also keep in mind that 'bright' or 'shimmery' sounds usually have many higher harmonics and 'dark' or 'round' sounds have fewer. Beyond that, you get to experiment and have fun - it's a great way to learn about the nature of sound. When using MIRAGEFREQ to help you when sampling, you can change the parameters on the Mirage as you need - they will be updated about once a second; this includes the sampling tone which can be turned on or off with the 'T' command. Additional commands allow you to set the number of waves per page anywhere from a quarter wave per page to four waves. When the loop size is more than one wave, subharmonics can occur. Check the Advanced Sampler's Guide for more on looping and loop size. While running MIRAGEFREQ, you may well notice small 'ticks' in the border of the 40 column screen. These are intentional and represent incoming MIDI bytes. This gives you some indication that the program is running and is also receiving data. I guess if there's one thing about this program you need to be aware of, it's that it can be a bit slow: I have used BASIC mostly because of the floating point calculations I've had to do (figuring out the internal filter frequencies that aren't listed was an amazing trip back down the road of logarithms and interpolation that I haven't been down in years!). Also, calculating the note number and error was interesting - you may want to check out the code to see how it was done! Anyway, using BASIC to calculate the values means that it takes about a second or so to loop through the program - the delay to your entering commands can take a little getting used to, but shouldn't be a problem - just stay relaxed and everything should be just fine - probably good advice in general. To change the waveform of the SID chip, edit line 171 (see the comments). U.K. users, and others that are using PAL based machines rather than the American based NTSC standard, will need to change the constant on line 215 as per the comments on that line. The reason is that these machines use a different crystal for the time base. I've checked the schematic and I think I have the right frequency, but since it isn't listed specifically, I've had to guess. If anyone finds a problem with the PAL constant, please let me know. (I'm sure that the value is at least close enough to be useful, even if it isn't 'exact'). Finally, there's the usual copyright notification and such... The MIRAGEFREQ program text and this accompanying MOREINFO text are Copyright Johnny Klonaris 1991 These programs are freeware. By that I mean that they are free to all who care to use them. All I ask is that if you distribute the software, you do one basic thing: Include all of the pieces unaltered: -miragefreq.bas -getrate.asm -moreinfo.txt *** OPINION ON *** Often, purveyors of software feel the need to express an opinion on the state of publicly distributed software: I am not above such things. Many who supply freeware feel the need to point out that they think shareware, where a 'donation' is encouraged, is wrong. I think this is a rather narrow view of the world. There is a place for both freeware and shareware. Authors often feel the need to prohibit others from making any sort of profit from the distribution of such software, even though such suppliers may well be providing a useful service. As the author of this program, I give you permission to make as much as your gonads will allow PROVIDED that you point out, by way of this message, that the program you are charging for is available free (if you know where to look). I feel the collection and dissemination of public domain software can be a valuable service and if done in a way that it can be supported by a market, then more power to them. So there! *** OPINION OFF *** In any case, if you'd like to contact me, there are a few ways. You can reach me on CompuServe at 75036,1351 though these days I rarely get on. You may get to me on Q-link as JawKnee. As of this writing, your best bet for reaching me via e-mail is on the Internet: johnny@cup.hp.com Those of you not connected to nor predisposed towards e-mail, can reach me via the United States Postal Service (a compulsory, non-competitive service) at the following address: Johnny Klonaris 2151 Old Oakland Road #371 San Jose, Ca 95131 Those of you on the Internet (this includes CompuServe members) can join, at no cost, the Mirage-Net. The Mirage-Net is an association of Mirage users connected by e-mail. For more information, send a message to: mirage-request@hpdsojk.hp.com on the Internet. From CompuServe: >INTERNET:mirage-request@hpdsojk.hp.com I'm interested in feedback. If you contact me, I'll keep your name on a list and will let you know about any future goodies that may be of interest. I hope this program helps you to make better samples with your Mirage. I'd love to hear the results (hint, hint!). Johnny Klonaris August, 1991 Addendum to Version 2.0 With revision 2.0, I've added the capability to use any MIDI sound source to generate a tone. New commands 'C' and 'B' were added for setting the channel and pitch bend range. Each command will prompt you for the required data (pitch bend range is in semi-tones or half-steps). Also added was the ' ' (space) command. Hitting the space bar toggles the MIDI note on and off. This should make sampling that killer synth sound a lot easier. Enjoy. Johnny Klonaris March, 1992 Addendum to Version 2.1 Version 2.1 fixes some bugs and handles error conditions much better. This version is intended to be referenced by an article in the Transoniq Hacker. Note: Version 2.0 was never released. Johnny Klonaris September, 1992