Unison Mode in SoundID VoiceAI

The AI-powered double-tracking feature in SoundID VoiceAI Perpetual Mode allows users to create natural-sounding double tracks from a single vocal track.

In this article:

• What is Unison mode in SoundID VoiceAI?
• What can you do with the new features?
  
  • “Offline controls” section
  • “Real-time controls” section
  • Width adjustment
  • Timing adjustment
• Preset selection
• Processing time and CPU usage
• Important to know

What is Unison mode in SoundID VoiceAI?

Starting with version 2.1, the Perpetual mode in SoundID VoiceAI introduces a new AI-powered approach to double tracking called Unison mode.

• Instead of requiring multiple takes, VoiceAI automatically duplicates the AI-generated voice to create up to eight double tracks in a single plugin instance.
• This allows for realistic variations in pitch and timing, simulating a true double-tracked performance.

What can you do with the new features?

• Create natural-sounding double tracks from a single vocal source (up to eight voices per plugin instance).
• Adjust pitch variations to introduce subtle differences between voices for a more realistic layered effect (Offline processing).
• Control timing shifts between voices using the Timing variance knob to create a more natural feel (real-time processing).
• Use the Width control to spread vocals across the stereo field or keep them centered (real-time processing).
• Enhance vocal production for music, post-production, and sound design.

Note: These features are available for the Perpetual license mode only (offline processing) and is not available for the Pay-As-You-Go mode. 

Controls

“Offline controls” section

• “Number of voices” spinbox – Sets the number of AI-generated double tracks (up to 8).
  
  
  
  
• Pitch variance – Introduces subtle pitch deviations for a more natural effect.
• Transpose – Adjusts the pitch of the generated doubles (already available in previous versions).
  * Processing settings do not apply in real-time and only take effect after processing is completed.

“Real-time controls” section

• Timing Variance – Introduces a slight timing shift between voices to make them sound more natural together.
• Width – Controls how wide or centered the generated voices are in the stereo field.
  * These controls work in real time, allowing users to tweak their double tracks instantly.

Note: if the Width feature is grayed out, it's likely due to the plugin being loaded on a mono track. To enable it, make sure that the plugin is loaded on a stereo input track instead. 

Width adjustment

• Setting the Width for Stereo Spread: Setting Width to 0 places all voices in the center, creating a mono sound. Increasing Width spreads the voices further apart, creating a wider stereo effect. 
• For an odd number of voices, the middle voice in the set will remain centered, while the rest are panned symmetrically. With three voices, the second voice is centered, with the first and third positioned left and right. With five voices, the third voice remains centered, with two voices positioned on either side. 
• For an even number of voices, they will be evenly distributed across the stereo field.
• When Width is set to 50, the voices are spread evenly between the left and right channels. At 100, they are pushed further apart, creating a wider stereo field.

Timing adjustment

• The Timing variance knob offsets the timing of each voice within a range of 2 to 50 milliseconds.
• Lower values result in smaller timing shifts. Higher values increase the timing differences between voices.
• The shifts are applied independently to each voice within the selected range.
• This adjustment allows for control over how tightly or loosely the voices align with each other.

VoiceAI double-tracking workflow

1. Capture the audio. Place a plugin on a track and capture your audio. Or you can copy the lead vocal part you want to double onto a new track and load a plugin there.
2. Choose the Processing settings. Set the number of voices (2-8). Adjust processing settings to add more randomness using the Pitch Variance knob (a recommended starting point is 30–40).
3. Start the processing - Multi-voice processing takes longer than single-voice processing.
4. Shape the sound with Real-time settings. Once the double-tracked vocals are set up and processed, the next step is to fine-tune their stereo placement and timing with Width adjustment and Time adjustment.
5. Finalizing the sound. After adjusting Width and Timing Variance, listen to the double-tracked vocals in the mix and make final refinements: Adjust the Width knob to ensure the spread fits the intended sound. Modify the Timing Variance knob to achieve the right balance between tightness and looseness.

Preset selection

Starting with this update, selecting a preset will no longer reset the positions of the knobs. If Auto Transpose is turned off, its status will remain unchanged after loading a different preset.

Processing time and CPU usage

The CPU usage is the same as for the perpetual mode with a single voice. The processing time takes longer as more voices are added.

Important to know

• There is a new plugin type: Mono > Stereo available in Logic Pro, Pro Tools, and other DAWs. This new plugin type is required for the Width knob to function.
• If you load the Mono version of the plugin (from previous versions), the Width knob will be disabled, and the plugin output will remain mono.
• The Real-time control section is hidden by default and will dynamically appear when the Voices knob is set to more than one.

Stereo -> mono plugin versions in Logic (left) and Pro Tools (right).