To answer Jeff's post, in fact, the SWAM virtual instruments _do_ have a way to derive attack intensity from the breath signal. You have to explicitly select "Expression" for the "Attack Sens." parameter. You can also fine-tune the response using the "Auto Expression", "Expr. Trigger Mode" and "Expr. Trigger Sens." parameters in the Options. It's not the default behavior because many SWAM users play them using a keyboard instead of a wind controller, but it's definitely there. Obviously, our onboard synth does this automatically on our sounds, when it makes sense.
I don't know about Jeff's setup, but there's no need to add MIDI velocity on top of the breath signal[1]. When a wind controller outputs a MIDI velocity, it's just something derived from the breath signal, but it's best that synths just use the breath signal and produce attack sounds themselves based on that. That's what SWAM instruments do when "Attack Sens." is set to "Expression", and I'd say it's the correct way. If your wind controller outputs a breath signal at a high enough frequency (the Sylphyo can do up to 1 kHz), then there's enough information for the synth to produce an attack that really sounds like what the performer intended to do on the controller. On the Sylphyo, you can observe this with the onboard sound Meow[2]. Attacks will sound different depending on whether your attacks are done using the tongue ("tu ku tu") or using the throat, but the processing stays minimal so that any advanced technique (flutter-tonguing or growling in the video) can be faithfully retranscribed, and it works well with a subtractive/FM sound like Meow (it might be more difficult for physical models).
[1] MIDI velocity could serve another purpose in a wind controller playing a SWAM instrument: telling how hard you're pressing a key when changing the note, similarly to how flautists sometimes use the sounds of their keys clicking in their performance.