I record my samples in Oremo with a condenser microphone, but when I import the voicebank in UTAU there's always this very nasally robotic engine sound on certain words (mostly "ra" and "na"). I tried fixing it by making the largest space possible between the consonant and the cutoff, I even tried deleting the .frq and regenerating it but it really doesn't help at all. When I record the samples, how long should I pronounce the consonants? What is the recommended tempo in recording VCVs?