I've seen some people talk about VOCALOID5 and whether or not it should provide backwards compatibility with VOCALOID4. People say yes, they'd like to use their singers in the same engine, and others say no, that it would be best for quality and improvement to cut pre-existing singers out of the picture. What do I think? Keep reading.
The original VOCALOID engine used a synthesis method similar to Moresampler, by Kanru Hua. It's based on a module of the voice not the actual samples, however, it wasn't as refined, obviously being from the early 2000s. VOCALOID2 completely switched its synthesis method to sample based and has kept that up until now. Some may say that it renders dead and dull voices, but this isn't exactly true. Uni's demos displayed a very vivid and lively quality, to which Dr. Yun said after Yamaha finished their touches on the voice, was lost. So is it really the method of synthesis, or is it the standards Yamaha describes as "quality"?
I bring this up because users argue VOCALOID should return to it's early synthesis method, which would do the same thing VOCALOID2 did, cut off backwards compatibility. Would this really be the answer though?
The transition to VOCALOID4 from VOCALOID3 is brought up as an example of how badly the engine needs to be updated, only adding two new additions and a new interface. However, the basis for this argument is completely void. The reason for VOCALOID4's lackluster improvement was not that it kept the same method of synthesis, it was that it was rushed by the, now former, president of the VOCALOID department.
I personally believe that we can keep backwards compatibility, to satisfy all users who wish to keep using their current and still improve, just like VOCALOID2 to VOCALOID3.