Comparison of the Yamaha WX5 and Akai EWI4000s

After about 20 years of playing the Yamaha WX series instruments I decided to try out the competition. The new EWI 4000 seemed to have the features I wanted, so I rented one at a local music store to check it out.

Despite being thrown off by the differences in fingering, octave keys and breath resistance, it was clear to me that the EWI had a nicer, more controlled 'feel'. It felt smoother and more under control than the WX5. At first I thought this might be an illusion created by the heavier, more solid metal body of the Akai instrument, so I decided to undertake some comparison tests.

Playing and recording the MIDI output for the same musical passage alternately on the two instruments I was amazed to discover that the EWI output a little over twice as much data for a given musical phrase as did the WX. Looking a little more closely I found that this was becuase it output nearly twice as much for the breath information and over THREE times as much for the pitchbend vibrato. The reason for this greater quantity of data seems to be that the EWI employs higher-resolution sensors for both breath pressure and pitchbend.

The graphs below illustrate the difference in breath sensor and pitchbend sensor quality between the Yamaha WX5 and EWI 4000s wind controllers.

The first four graphs show the same two bars of a Handel Flute Sonata perfomed with both the WX5 (left) and EWI4000 (right).

The second group of four graphs show the data for a smooth crescendo-decrescendo (16 beats long) and for a single measure of steady vibrato.

These graphs were made using MIDIForth, my Forth-based MIDI editing system.

WX5 Lip sensor It's hard to make a nice vibrato on the WX5! This is an attempt to keep the pitch 'centered' (the middle horizontal line indicates 'true' pitch)	EWI4000 lip sensor Note the regularity and high data density (about 3x the WX5). It's also centered on 'true' pitch (the '0' line)
WX5 breath sensor Note the difference in general data density. On gently sloping segments only about 1/2 the possible MIDI values are filled in.	EWI 4000 breath sensor The data density is about 2x the WX5. On gently sloping segments all possible MIDI values are filled in.
WX5 breath resolution Note the large jumps in BC value ("DATA 2"). Even smoothly sloping segments change by twos.	EWI breath resolution Note the nice smooth series of consecutive DATA 2 values.
WX5 Long Crescendo This 16-beat crescendo-decrescendo demonstrates the difference in data density between the two instruments. Thee WX5 put out 199 BC points to fill in this long note...	EWI Long Crescendo ... while the EWI4000 put out 347, nearly twice as many. The difference of data quality is evident in the more 'solid' 'filled-in' look of the graph.
WX5 Flat Vibrato Many windlisters have complained about the WX5 "flat-spot" or inability to create a nice vibrato. I've been battling with it for years. This graph illustrates a couple of the problems. First, it's almost impossible to do a 'centered' vibrato. The pitch tends to vibrate FLAT of true pitch, giving the overall note a flat quality.	EWI Centered Vibrato The EWI's 'bite-blip" vibrato system is intrinsically 'centered'. The resulting vibrato is quite 'sweet' by comparison to the 'flat' WX5. It has the disadvantage of being unable to do pitch 'scoops' or 'slides' however. Note the big difference in data density - the WX5 bar of vibrato consists of 136 events, while the EWI's consists of 472 - about 3x as many.