DeepVocal Wikia
Advertisement

Bukimi (不気味) is a feminine androgynous vocal library for the upcoming DeepVocal synthesis engine, developed by peYton.

History[]

2020[]

Development of what is described as a port of Bukimi's English voicebank over from UTAU to DeepVocal began in late February of 2020. Given the current state of the phonetic system of DeepVocal, Bukimi's port has been described as frustrating to work with.

A very early build of Bukimi's voicebank that can be tested can be downloaded here.

As a personal reminder from peYton himself, please don't expect a very HQ voicebank or a final product. This is more like a proof of concept voicebank that exists for the purpose of testing things out!

2021[]

Bukimi's samples have been normalized and two more pitches have been added to their voicebank! They're now 3 pitches in total!

The new voicebank can be downloaded here.

Phonemes[]

Vowels[]

  • [a] = open front unrounded vowel
  • [i] = closed front unrounded vowel
  • [u] = closed back unrounded vowel
  • [A] = open back unrounded vowel
  • [I] = near-close front unrounded vowel
  • [U] = near-close back rounded vowel
  • [E] = open-mid front unrounded vowel
  • [V] = open-mid back unrounded vowel
  • [O] = open-mid back rounded vowel
  • [aI]
  • [aU]
  • [eI]
  • [OI]
  • [oU]
  • [a3]
  • [I3]
  • [E3]
  • [O3]
  • [@] = schwa
  • [@l] = [l] colored schwa
  • [3] = open-mid central unrounded vowel
  • [{] = near-open front unrounded vowel

Consonants[]

  • [b] = voiced bilabial plosive
  • [d] = voiced alveolar plosive
  • [f] = voiceless labiodental fricative
  • [g] = voiced velar plosive
  • [h] = voiceless glottal fricative
  • [j] = palatal approximant
  • [k] = voiceless velar plosive
  • [l] = alveolar lateral approximant
  • [m] = bilabial nasal
  • [n] = alveolar nasal
  • [p] = voiceless bilabial plosive
  • [r] = alveolar approximant
  • [s] = voiceless alveolar fricative
  • [t] = voiceless alveolar plosive
  • [v] = voiced labiodental fricative
  • [w] = labial-velar approximant
  • [z] = voiced alveolar fricative
  • [4] = alveolar flap
  • [tS] = voiceless postalveolar affricate
  • [D] = voiced dental fricative
  • [dZ] = voiced postalveolar affricate
  • [N] = velar nasal
  • [S] = voiceless postalveolar fricative
  • [T] = voiceless dental fricative
  • [Z] = voiced postalveolar fricative
  • [ts]
  • [sk]
  • [sm]
  • [sn]
  • [sp]
  • [st]

Concept[]

They are a Bat.

Bukimi was originally intended to be some sort of weird, distorted, shouty voice that was in no way to be taken seriously, but over time, they slowly became much more than that.

Bukimi's real name is "Christine" and is a bat. Their scarf functions like wings and/or external appendages, and typically follow arm movement by default. They're clumsy and constantly getting hurt, indicated by all the stitches and band-aids. Bukimi has a black eye that will never heal, and only half of their mouth functions because of a really bad injury during their childhood.

Their bowler hat is high-tech and is arguably sentient. Its name is currently unknown.

Etymology[]

  • 不気味 - Bukimi; Japanese word for "eerie"
  • Christine - (first name); common English feminine name
  • Quaranstein - (surname); Quarantine + Frankenstein

Appearance[]

  • Fur color - Sometimes depicted as jet black, however a pastel purple fur pattern is more common
  • Eye color - Purple.
  • Outfit - Striped scarf, white tank top, and either a skirt or booty shorts.
  • Nail colors - Typically painted different shades of purple, but they can be depicted with any color nails.

Product Information[]

Product Information
Optimum
  Genre: Pop, Indie  Tempo: 80-150  Range: G3-D5
Basic Information
Bukimi has a soft and breathy androgynous tone.
  • They have three pitches recorded in the newest build.
Bank/Phonetic Information
  • X-SAMPA (Peiton style)
Software Issues as noted:
  • There are some odd buzzing sounds that may occur unless the breath parameter is turned down very low.
  • There are no vowel transitions as of build 0.1. Some other transitions may also possibly be missing.
  • Since consonant transitions aren't natively supported, [C0] notes should be placed before any unvoiced consonant transition. For example, [last] should be inputted as [l{][s0][st][t].
Voicebank Sample

Build 1:

Build 2:

Demonstrations

Demo 1 - Rotary Dial SoundCloud
Demo 2 - I Am Not A Robot YouTube SoundCloud

Gallery[]

References[]


External links[]

Advertisement