User blog comment:Serza5/Ideal Append/Tone System/@comment-74.99.129.105-20141103010633

The reason appends and extra voicebanks exist is because that is not a viable way to add expressions.

Using parameters or a parameter-like function to alternate or gradually shift between voicebanks wouldn't work because:

A. Depending on the different tone of voice, the consonants and vowels would be at different lengths, which wouldn't line up as the parameters changed

B. The pitchsets wouldn't match up with each other because it's harder to perform certain expressions at different pitches (try whispering up high or whistling down low)

C. As the parameters changed, the wavelengths of the two voices would overlap and cause distortion since both voices wouldn't be on the exact same pitch.

We already have some voicebanks that do their best with creating an expressive vocal in one library (like how Kokone sounds vastly different in the treble, bass, and falsetto ranges).

On a related note, v.Connect STAND (also known as Cadencii), does support what you're talking about. You an create a VB that uses morphing technology to add BRI or BRE to an exsiting voice by adding in your own BRE or BRI samples. But the results are very spotty and come nowhere near Vocaloid.