Creating quality ring tones is less about mobile phone technology than it is about music composition. When our tones are created we cannot merely take a short section of an MP3 tune, run it through a file conversion process, and expect to come up with competitive content.

There are frequency limitations when composing ring tones. We need to consider the general register of the device and also understand the polyphony level, or number of simultaneous voices, available on each handset. For example, Nokia 7210 and Nokia 3510 handsets support four simultaneous tones (polyphony level 4), while Nokia 7650 has a polyphony level of 24.

Four-voice phones are capable of producing a full sound, and there is a genuine difference between monophonic SMS ring tones and [even four-voice] polyphonic MIDI ring tones.
In addition to different levels of polyphony, different phones produce their sound in different ways, and therefore require different file types. Some mobile devices use software synthesis, for example, using the Beatnik Audio Engine, where others use a hardware or proprietary system to do so.
Nokia devices that support polyphony use software synthesis via the Beatnik Audio Engine and support a subset of the MIDI file format. The same audio engine (and therefore same file type) is used in some Sony phones and the Danger Hiptop.

Some other polyphonic devices require the SMAF format.SP-MIDI (Scalable Polyphony) is a new file format for ring tones and is a subset of General MIDI. It is also distinct from GM-Lite, which requires a 16-voice hardware chip. SP-MIDI will actually run on a full MIDI.

Synthesizer such as the Beatnik Audio Engine is flexible enough to allow the same file to be played on devices with a range of polyphony levels. Additionally, it takes an elegant approach to degradation and also allows full control over phone vibration.

The key to making good ring tones is in the preparation. Tracks are organised according to channel priority and then mute and unmute various tracks to simulate behavior (note stealing) in devices of a different polyphonic level. The data is also optimized by analyzing overlapping notes and possibly eliminating the use of sustain, as different devices are known to sustained notes differently.

When composing or recomposing an existing piece of music, we zero in on the most important part of the tune to use in the ring tone so that the tune can be recognised in the finished song.