Skip to content

USER_GUIDE.md

Latest commit

 

History

History
68 lines (48 loc) · 3.53 KB

USER_GUIDE.md

File metadata and controls

68 lines (48 loc) · 3.53 KB
Raw

FlowReVoice User Guide

Transform a vocal

  1. Insert FlowReVoice on an isolated vocal track.
  2. Put it early in the effect chain to capture the raw voice. Effects placed before it become part of the capture.
  3. Select the Applio root, .pth model, and optional .index.
  4. Move the DAW playhead shortly before the wanted vocal.
  5. Press Arm Capture.
  6. Press Play in the DAW and let the section play once.
  7. Stop playback. Conversion begins automatically when Transform when playback stops is enabled; otherwise press Transform.
  8. Wait for the status to become Voice Ready.
  9. Rewind and play the same section to hear the converted voice at its original position.
  10. Press Export AI WAV to create a normal 24-bit WAV for the DAW playlist.

The first playback pass stays transparent. Applio conversion never runs on the real-time audio thread.

Change the voice

Select another .pth and optional .index, then press Transform again. FlowReVoice retains and reuses the untouched capture rather than converting its own AI output.

Controls

  • Model Range: Semitone shift before AI conversion. Use it to match the source singer to the training voice range.
  • Accent: Influence of the optional .index file.
  • Consonants: Protection for unvoiced consonants and articulation.
  • Dynamics: How strongly the output follows the input amplitude envelope.
  • Pitch Lock: Applio pitch-stabilization strength when its toggle is enabled.
  • Output Pitch: Independent pitch shift after AI conversion when Post pitch is enabled.
  • AI Mix: Converted/original blend. At 100%, the captured section is a true AI-only replacement.
  • Phrase splitting: Separates long audio into vocal phrases and keeps every phrase independently anchored.
  • Output cleanup: Uses Applio's output-cleaning stage.

Balanced, Detail, and Character are starting points and never replace the chosen model or pitch range.

Save and load presets

Press Save Preset to store every control, toggle, pitch setting, detector choice, and selected Applio/model/index path as .flowrevoice.json. Press Load Preset to recall it.

Presets do not contain audio or model data. If a preset came from another computer, relink paths that do not exist locally.

Export behavior

FlowReVoice exports a timeline-length 24-bit WAV representing the captured section. Silence and phrase spacing are retained, so the export lines up when placed at the original capture start.

Cache locations

Raw captures and AI results are stored locally under:

%APPDATA%\Flow Studio\FlowReVoice Cache

The version-adaptive Applio bridge is stored under:

%APPDATA%\Flow Studio\FlowReVoice

Troubleshooting

  • Plug-in is missing: Confirm it exists under C:\Program Files\Common Files\VST3, then run a complete plug-in rescan.
  • Nothing captured: Arm before pressing Play, confirm audio reaches the insert, and keep the track unmuted upstream.
  • Capture stops after seeking or looping: This is intentional; timeline jumps end capture so unrelated locations are not joined.
  • Applio warning: Select the exact folder containing core.py and env\python.exe.
  • The result is not heard: Rewind to the same timeline location after Voice Ready or export the AI WAV.
  • The beat changes voice: Remove FlowReVoice from the master or bus and place it only on the vocal track.
  • First conversion is slower: Applio must load Python, AI dependencies, and the model. GPU capability, model size, and capture length affect processing time.