GETTING THE MOST OUT OF YOUR SAMPLER By MIKE ARCHER. Edited by PARASITE. There's no doubt that the quality of the samples you make on your Amiga is very dependant on the sound sampler you are using, however, I'm not going to write an article on the quality of various sound samplers, mainly because I have only ever used one sampler on the Amiga. (Any firms out there who want to send me a free sampler to review are welcome to!!) No matter which sampler you own, there are quite a few ways of improving the quality of the sample it produces. Most manuals that come with samplers merely explain how to operate the features of the sampling package provided, and not the correct way of going about making a sample of clear quality. Following are some hints and tips on how to reduce distortion and background noise... INPUT VOLUME Many sampling packages provide a feature to view the waveform in real-time like an oscilloscope. This feature isn't just a fancy graphic equalizer to rig up to your hi-fi (Though it does make a damn good one!) You should always use this display before making a sample to adjust the volume of the sound you are sampling. The volume should be set so that the waveform shown just reaches the top and bottom of the display, but doesn't go off the display at all. If your input volume is too loud then when you sample the sound the top & bottom peaks of the sound wave will be cut off, this causes very bad distortion. If your input volume is quiet then the background noise produced by your sampler will be heard. To my knowledge all samplers currently available produce some background noise while sampling. The louder the volume you sample at the more this background noise is drowned out. If you are sampling a quiet instrument, you should still adjust the input volume as above, then use the volume controls in the music package you are using to make that sample quieter than the rest. CUTTING OUT THE INSTRUMENT Once you have made a sample, and you come to cut out the part of the sample you wish to save and use as an instrument in your music program, you should always ensure that the wave at the start and end of the sample are both zero. In other words, always cut out the sample from/to a point where the waveform crosses the center line. Doing this will vastly reduce any clicking noises that occur when a sample starts/finishes playing. SAMPLE RATE The sample rate you use is really up to you, the faster the sample rate the clearer the samples produced will be. But they will also use up more memory, it's really up to you to decide how much quality you are willing to sacrifice to save memory. If you are sampling a musical instrument that you intend to use along with other samples of other instruments, it is very important that your sample is in tune. The easiest way to ensure this is to use a sample rate which is the same frequency as the note you are sampling. Below is a table of what sample rates match up to what notes. For example, if you are using a sample rate of 338, you should play an 'E' on the instrument. NOTE SAMPLE RATE FREQUENCY (Hz) Middle C 427 261.7 C# 404 277.2 D 381 293.7 D# 359 311.2 E 339 329.7 F 320 349.3 F# 302 370.0 G 285 392.0 G# 269 415.3 A 254 440.0 A# 240 466.2 B 226 493.9 C 214 523.3 If you wish to sample using rates not on this table then the table can be easily extended by the following rule... For each octave higher you go the sample rate should be halved and the frequency doubled. SAMPLING FOR REMIXING If you wish to remix a song by sampling short chunks of music and piecing them together, you will probably discover that it is very difficult to make it so a sample fits in with the tempo of the music program. Usually your sample will end playing slightly before/after the next note is played. (the next note usually being the next sampled bar of music.) This will cause either a very noticeable click or a jump in the rhythm, depending on how far out it is. The only way of getting round this problem is to slow down or speed up the sample ever so slightly when it is played back. Unfortunately most music packages will not support doing this, so we have to get around the problem another way. If you record a sample slightly slower than the rate you play it back at, this will effectively speed the sample up slightly. You could sit there resampling the music over and over again at slightly different rates until it fitted the tempo of your music program. This is what I used to do, how- ever I decided there must be an easier way so I worked out a little formula for use with sound tracker (and other clones). n = ( tempo * 10000 ) / ( rate * 1.396825 ) TEMPO - This is the tempo setting you will be using in Sound Tracker. RATE - This is sample rate which you will play the sample back at, use the table given before to work the corresponding note to be played. The value 'n' returned by this formula will be how many bytes of sample are needed to fit exactly one note. So you must multiply this value by how many notes your sample will last for. This will then give how many bytes long your sample should be. Then it is a case of resampling at different rates until the sample of the music is the required number of bytes. OK, I admit this still involves alot a messing around trying to find the right sample rate, but at least you do not have to keep saving your sample and trying it in Sound Tracker to see if it fits!