Last month, I took a broad view and considered ringtone development in the context of the audio qualities of the handset and the requirements of the end user - it seemed to me like a reasonable starting point. In this article, we'll focus on the MA chip itself head straight for one of the most important features of the MA sound engine, and that is the onboard FM synthesizer. It's worth noting that having a parametric synthesizer (that is, a synthesizer with variable controls) embedded in a consumer device that is sold to millions of people, is a real milestone in the history of electronic music!

What are the benefits of the FM synth in the MA audio engine? Well, there are two very important ones:

1. You can create unique voices, purpose built for the Ringtone - the MA FM synth has over 70 user programmable controls for creating and modifying sounds.
2. The clarity of the FM tones is excellent - the audio output is "calculated" in realtime from the set of FM parameters and does not require large amounts of data to be stored and read, like wavetable synthesis. This means it can be rendered at a very high quality, in fact the sampling rate for the FM output is 48Khz , theoretically even better than a CD.

For some developers however, FM synthesis seems to be complex and difficult to manage, and consequently they don't really get to enjoy these benefits to the full. FM is actually a very straightforward synthesis method, with a simple set of basic controls - though it can look a bit strange to those programmers who are more used to filters and wavetable systems. I'm guessing that for these folks, an important question is "How do I quickly modify an FM voice; make it brighter or less metallic or smoother and so on?"

In this article I would like to answer this question by suggesting an intuitive way of looking at FM in order to familiarize it, and then offer some simple practical exercises of voice tweaking. Next month we'll strengthen our knowledge by taking a look at the underlying theory of FM and do some ear training exercises. Together, this will reduce the mystery of FM so that you can feel comfortable with it and enjoy applying the benefits to your ringtones.

A good way to get a handle on FM is to break the synthesis model down into separate audio "chunks" that contribute some element of sound to the final voice. Once these chunks have been identified by looking at the algorithm, they can be simply turned up or down, on or off, to see what they do. In order to do this, you need to understand how the sound is built up in the algorithm, and how sound appears to "flow" through it. Have a look at this "algorithm" diagram, which shows how the operators are connected...

First of all, each operator is basically a very simple oscillator, capable of making a sine wave (other waves are possible, but for the moment, let's keep it simple). It is quite reasonable to think of sound as flowing from left to right out of each operator, through the connecting pipes, then "brightening" - that's the FM process at work - the output of every operator that it flows through, until it arrives at the output on the far right. The only way for sound to come out of the system is "through" an operator. There are two names, or states, for operators - they are called "carriers" when they are connected directly to the audio output, or they are called modulators when they are connected to the input of another operator. When the output of any operator is OFF or set to minimum, no sound flows from it or through it, and this is how you can easily listen to the effects of individual operators. In the case of the operator configuration, or Algorithm, shown in the diagram, its behaviour can be described like this...

• Operator 4 outputs a sound - it is connected directly to the audio output. For the algorithm shown, all sound flows through this operator therefore its output level and envelope are "master" volume controls.
• Sound from operator 3 flows into 4, modifying its output and making it sound brighter. The output level and envelope of operator 3 define how much sound flows from operator 3 into operator 4.
• Likewise, the sound from operator 3 is already modified and brightened by the sound from operator 2 flowing into it.
• Along the top of the algorithm, sound from operator 1 joins the flow from operator 3 into operator 4, adding its own effect on operator 4 along with that of operator 3.
• And finally, the little square loop around operator 1 represents "feedback". The sound from operator 1 is being spun back into itself (just a if there were a clone of itself further upstream) before passing along the pipe and modifying the sound of operator 4.

Here are two other algorithm diagrams. Spend a moment to think about the sound flow in each one, and ask yourself how each operator might affect the sound.

Question: what happens in each case when Operator 3 is turned off?

Answer: In the case of Algorithm 4, Diagram 3, operator 4 would sound but it would not be modified at all, so if it was set to be a sine wave, that's all you'd hear. For Algorithm 5 shown in Diagram 2, the operator pair 3 & 4 would become silent, but the complete operator pair 1 & 2 would continue to sound, the sound of operator 2 still being modified by operator 1.

With this little piece of information and insight, you can now identify and explore the audio "chunks" of any FM voice by simply by adjusting the output level of different operators. You'll be surprised how quickly a little exploration in this fashion will give you a feel for controlling FM voices, and you can actually modify sounds quite effectively, using only a few of controls:

1. the output level for carriers (TL) - affects the volume of its particular chunk of the algorithm
2. the output level of modulators - affects the brightness of the operators to to which it is connected
3. the attack time (AR) and release time (RR) of carriers - affects the shape of the volume of the sound
4. the attack time and release time of modulators - affects the shape of the "tone" of the sound

Download the MA-3 Voice Bank DavesWebExample2.vm3, and when you have done this, open the voice file into the SMAF software tool. Double click on voice "FngrBass/SC", the first patch in user bank 1, and explore it using these controls. Diagram 4 shows how to get to the controls you need. When you've done that have a go at the second voice, which uses a different algorithm.

It's not so difficult is it! With this approach you should soon be able to make those little tweaks to FM voices so that they really work well with your tracks. And don't forget, if you have any questions, we'll be glad to answer them for you, through the forum. Next month, we'll take a quick look at the underlying FM theory and see what actually goes on between the operators.