EPS/ASR Sample Rates

Question: I have a EPS 16-Plus, When pressing the effect button, one of the options is playback at 78khz. Imported a 16 bit 88 khz sound with a lot of high frequencies from my PC and it played the high frequencies very accurately. Does it mean that 16-Plus supports playback of sample rates greater then 44.1? Can ASR do that? Can the Waveboy effects be ran in this mode?

Answer: Let's get an overview on how the EPS/ASR does sample rate recording and playback.

A stored sample - a .wav file, AIFF, an Ensoniq sample on disk - all have the same properties. They usually contain a set of numbers ranging from 0 to 65535, which correspond to a volume level at a very very short period of time. That period of time is called the SAMPLE RATE. A common sample rate is 44.1kHz. That is what CD's are stored at. That means that the playback engine that plays the sample back is required to process 44,100 numbers in exactly a second's time.

A CD player is extremely simple in that regard. CD's are standardized at 44kHz playback, and any CD player simply plays the numbers back at that rate.

But, perhaps you have seen some CD players that offer pitch control. All they do is change the playback rate slightly depending on where you turn the pitch control knob. Changing the sample rate changes the pitch of the sound - higher sample rates raise the pitch, lower sample rates lower the pitch.

Now, first complication: most modern CD players actually DO NOT play the sound back at 44,1000 numbers per second. Have you seen the word INTERPOLATION floating around the documentation? This is how this works: the CD player reads a set of numbers. It then inserts numbers in between each number mathematically, thus making the sound more precise and clear. Then they change the playback rate. For example, if a CD player has 2x interpolation, it inserts one number between each 32 numbers, and plays back the sample at 88.2kHz. That way the pitch is not affected.

Now, to the EPS/ASR. This is the INTERPOLATION concept on steroids. Let's take one concept at a time, and let's take the 16-Plus as our example.

Let's say you have a 44.1kHz sample (that is, recorded at 44.1kHz), with the ROOT KEY at C4, and the effect being HALL REVERB, which has a playback rate of 30kHz. When you press C4, this is what the 16-Plus does:

That's a lot of processing!

(By the way, the playback rate, even when set to 30kHz, is not REALLY 30kHz. The EPS/ASR interpolates on playback 8x.)

When you play a B3 note, the 16-Plus has to take the extra step of adjusting the sample rate to reflect the half-step pitch difference. Now imagine playing several notes at once. Now imagine playing a lot of notes very quickly. Wow!

In other words, the EPS/ASR sampler families use extreme sample rate adjustment to simulate a multi-pitch keyboard.

The 16-Plus has effects that playback at 30kHZ and 44.1kHz, and non-effect playback rates of 30k, 44.1kHz, and 78kHz. Faster playback sample rates put greater stress on the CPU/sound engine chip, and that's the reason that a choice of a higher playback rate results in less voices being played back polyphonically. That's the trade off; using higher sample playback rates improve the quality, but decrease the polyphony.

Effects are tied into the sample rate issue because that's part of the digital output scheme.

That's pretty much it. One last complication: changing the ROOT KEY essentially changes where the EPS/ASR knows to play back the sample without any adjustments BESDIES the one needed to get the data to the correct playback sample rate. However, you can change the rate listed as the recorded sample rate too. (In other words, you can lie.) You get to this by pressing COMAND-WAVE-WAVESAMPLE INFORMATION. Go to the SAMPLE RATE: parameter, and you can change it! Now, this doesn't improve the sound quality - it just changes the reference point the EPS/ASR uses to playback the sample. If you use the CONVERT SAMPLE RATE function, this actually changes the sample data to reflect the new rate. You can call this "hard-interpolation." This may or may not improve the fidelity of the sample - the computer is at best guessing what the correct values should be.

Contributed by: Garth Hjelte

Back to EPS and ASR-10 Knowledge Base Index

Close Window