Introducing Synthesizer V,
the Revolutionary Singing Synthesizer
Science for Art's Sake
We aim to create a tool that ultimately leads to the artistic perfection of artificial voices, and the passion has been put into a 7-year quest for the scientific modeling of singing voice at the highest calibre. Synthesizer V, coming straight from the laboratory, is the outcome at the fifth iteration of this research project.
The Synthesizer V Engine
Based on a hybrid of artificial neural networks and concatenative synthesis, Synthesizer V delivers natural voice from a small amount of sampled data. Our patent-pending Low Level Speech Model (LLSM) separately processes vocal folds and vocal tract features, thereby allowing for high-fidelity and flexible manipulation of voice timbre.
Maximal Working Area
The editor was designed under the guiding principle of maximizing work efficiency while keeping things simple. We dedicated the entire window to the piano roll with minimizable panels overlayed on the top for auxiliary information. Semi-transparent design allows panels to be seen-through, revealing the notes underneath.
Real-time Pitch Visualization
Pitch bends can be edited either in the pop-up panel on a note-by-note basis, or as one of the continuous parameters alongside tension, breathiness, etc. The editor visualizes the pitch curve (including vibratos) and updates the pitch preview display in real-time.
Smart Lyrics Editing
The built-in natural language processing module understands both words and phonetic symbols. For Chinese and Japanese, lyrics can be keyed-in as either glyphs or romanized spellings.
Synchronized Parameter Editing
Under "Sync-Parameter Mode", the changes to notes are automatically broadcasted onto the parameters below. This allows one to drag-drop, copy and paste parameters along with notes, greatly simplifying the workflow.
Key-bindings for all editing commands can be remapped at your preference. The parameter panel can be detached from the bottom to align with the notes. The editor also comes with several color schemes for you to pick.
Synthesizer V sings in US English, Japanese and Mandarin Chinese with plan for more languages and dialects; the editor has been translated to Japanese, Chinese and Korean.
Synthesizer V as a platform strives for supplying and supporting music producers with the most flexible and high-quality singing synthesis technology. Feedback from singers, artists and businesses with an interest is welcome. This technical preview showcases a part of our outcomes and we look forward to your comments and suggestions.