` Keiro

shikeii

Keiro

The first Russian male AI singer

CV: keirokeer

Download
Voice Model
How to
Install

Катюша (Katyusha)


Video demonstrations









Information



  • Native language Russian
  • Additional languages (fully supported)Japanese, English, French, Korean, Thai, Chinese
  • Additional languages (indirectly supported)Spanish, Italian, Portuguese, German, Ukrainian, Arabic
  • Vocal modes Standard, Soulful, Black, Haze, Vampiric, Kinetic, Sharp, Growl
  • Supported expressions Tension (curve), Velocity (curve), Gender (curve), Pitch Expressiveness (curve)
  • Voice provider keirokeer
  • Production keirokeer, mii-chandesu, serverpodivan
  • Data size 10+ hours
  • Data used for multi-language Chikano1582, Kungom, PRINTmov, Tigermeat, Miyo, M4Singer, Mim, Gyromancy, VictorVonDrew
  • Additional assistance Tigermeat, PixPrucer

Multi-language demonstration



Japanese

Ballade by Eiji Hirasawa

English

Twinkle, Twinkle, Little Star

French

Synthimentale by Conslo

UST by UFR

Korean

Boundary by Dr. Yun

UST by Arctic Feast

Chinese

茉莉花 (Mo Li Hua)

Thai

รักแรก (First Love) by NONT TANONT

UST by Tempoprimo

Spanish

Me Cuesta Tanto Olvidarte by Mecano

UST by ryusouta



Available Vocal Modes

Standard

Normal tone, used as a base for the singer

Soulful

Airy, relaxed vocals

Black

Normal tone used until 2023. Slightly worse, but has its own peculiarities in tone

Haze

Whisper or ASMR. Not a great experiment, but can be used in specific situations

Vampiric

Relaxed soft vocals with manic tones and a sad timbre

Kinetic

Adult, strong vocals. Good for rapping because made from speaking data. Voice acting

Sharp

Strong, powerful vocals

Growl

Can be mixed with another strong vocal mode for a growl effect


Available Parameters

CLR: voice color

Allows you to switch between vocal modes of the singer. You can also add parameters for each vocal mode to mix them between each other (see Features)

TENC: tension (curve)

Allows you to change the strength of voice by drawing it with a curve. This can greatly emphasize the vocal modes of singer and make the vocals more expressive

VELC: velocity (curve)

Makes consonants clearer during fast vocals. If singer stutters, increasing the values of this parameter will improve pronunciation

PEXP: pitch expressiveness (curve)

Lowering the values of this parameter will make the pitch generation less expressive

GEN: gender (curve)

Allows you to change the vocal formant

DYN: dynamics (curve)

Allows you to change the vocal dynamics


Artworks


OtomeHime

shikeii

OtomeHime


Yonanta

TheLeonsDoodles

OtomeHime


shikeii

Tenshi

OtomeHime


OtomeHime

LOP-P

OtomeHime


shikeii

powimo

OtomeHime