Transform connected source audio toward a connected or selected target voice while retaining a generated audio result for preview and download.
Where: Flowgen → Add → Audio → Voice Swap
Before you start
- Have permission to use both the source recording and target voice.
Key ideas
- Source plus target: Connect Source Audio and either wire a Voice Profile or select the target voice in the node.
- Provider/model settings: The inspector exposes provider/model and supported voice-similarity controls such as stability, similarity, and noise handling.
- Audio result: The swapped audio is previewable on the card, downloadable, and wireable into lip sync, composition, or audio render.
- Voice Swap card: VoiceSwapNode uses the generic shared-card path. Transform source audio toward a selected or connected target voice profile.
- Properties sidebar: renderAudioNodes.tsx
- Opened surfaces: Audio preview; download
Steps
- Connect the source audio and choose or wire an authorized target voice.
- Select the provider/model and tune only the settings it exposes.
- Run the swap and compare intelligibility, timing, background noise, identity similarity, and artifacts.
- Download the result or connect it to Lip Sync, Audio Render, or the video timeline.
Voice Swap controls and presets
- Add path: Flowgen → Add → Sound → Voice Swap.
- Card: VoiceSwapNode.
- Properties: renderAudioNodes.tsx.
- Controls: Provider/model; source; target voice; stability; similarity; noise.
- Opened surfaces: Audio preview; download.
Tips
- Clean the source before swapping when noise or overlapping speakers confuse the model.
- Keep an unmodified source for comparison.
Limitations and important notes
- Similarity controls cannot guarantee identity, pronunciation, emotion, or artifact-free output.
Troubleshooting
The result sounds unstable or noisy
Use a cleaner single-speaker source, reduce extreme similarity settings, or try a different supported model/provider route.