Image by JaimeSynth

Usage Examples

Vocal Mode- Blossom

Vocal Mode- Petal

Vocal Mode- Lush

Vocal Mode- Uprooted

Vocal Mode- Soil

Vocal Mode- Flourish

Vocal Mode- Komorebi

• Bluesky
• YouTube
• Discord

About

Leif is a vocalist designed for the AI vocal synthesis program Diffsinger on OpenUTAU, made in collaboration with JaimeSynth and TigerMeat. He is recorded by JaimeSynth in English and Japanese, however has support for many other languages like Spanish, French, Italian, Portuguese, Chinese, and Korean to name a few. With a dataset of ~132 minutes, roughly equally split between six unique vocal modes, he can provide vocals of good quality with a bouquet of different vocal stylings

Bio

Etymology- The Name “Leif” is Norse for heir, as his appearance and voice style match that of his VA’s. It is also a play on words as his entire outer appearance is comprised of leaves and various floraOrigin- Leif is a plant-based organism that has developed sentience over the course of human evolution. He may have started as sentient fungi, or perhaps a flower? His true genesis is up for debateAppearance- The flower atop his head can change in size and shape at will. He also has the ability to alter the texture, size, and shape of his body’s foliage, allowing his form to fit whatever serves his needs. His eyes are just larger versions of the petals to the flower in his hair, hence there being no scleraPersonality- Being quite literally nothing but a mass of plants, he can seem quite aloof or unfocused. His behavior, while vaguely human, can seem awkward and unusual. He does go out of his way however to display acts of kindness and cares deeply for those in close proximity to him. Depending on his mood, he can sprout different colors of flowers in his beard. Being comprised of nothing but plants, Leif technically does not have a gender, but uses he/ they pronouns and seems to be attracted to men

Vocal Modes

Blossom (21.98 minutes) - Bright and cheery vocal mode, recorded with a constant smile on the VP's face (which was quite exhausting)Petal (21.86 minutes) - Soft-spoken, light vocal modeLush (21.96 minutes) - Rich, deeper, smooth sounding voiceUprooted (21.44 minutes) - Sharp, rough voice made primarily for high energy songsSoil (22.18 minutes) - Deep, low, nearly gravelly sounding voiceFlourish (_{22 minutes) - Vibrant, dynamic voice made for songs that may require some power. Recorded solely in English}Komorebi (24 minutes) - Enka inspired voice with lower pitches akin to Soil, and higher notes with similar power to Flourish but more controlled. Recorded solely in Japanese

Download

• Leif Diffsinger

• Leif UTAU (Original CV Ver)

About Diffsinger

Diffsinger is an open-source AI-powered vocal synthesis software commonly used within the freeware platform OpenUTAU with a broad variety of vocalists largely provided by fans of vocal synthesis. While the program uses AI to produce realistic vocals, it requires human skill and artistic style to produce said vocals and make them sing. The vocalists are produced by having a human sing training data, meticulously labelling the phonemes within the recorded audio, then having a computer train on that data to grab the characteristics of the voice. Using the software is as easy as laying out notes on a midi keyboard, adding lyrics, and editing thin tuning lines and parameters to achieve the desired soundThe precursor to this technology, concatenative synthesis, which similarly uses recorded and labelled raw vocal samples but has the raw samples rendered directly into the software with labels for timing and smooth transitions, has been around since the 2000s with examples like Vocaloid and UTAU. Diffsinger works similarly to the AI vocalist product Synthesizer V, but is typically free to use and has the added benefit of allowing one to create a vocalist from their own voice