Back to blog
Tips & Tricksoptimizationpitch shiftvoice strengthfine-tuning

Optimizing Pitch and Voice Strength for Perfect Covers

Master the two most important settings in OG Voice: Pitch and Voice Strength. Learn how to fine-tune your generations for maximum realism.

OG Voice TeamMarch 2, 20262 min read

The Secret Sauce of Generation

When you use the Generate form on OG Voice, you'll see two primary sliders: Pitch and Voice Strength. While the AI defaults are good, understanding how to adjust them can take your cover from "good" to "unbelievable."

1. Mastering the Pitch Slider (-24 to +24)

The Pitch slider shifts the voice in semitones.

  • 0: No shift. Orijinal vocal frequency remains as is.
  • +12 / -12: Shift up or down by a full octave. Perfect for gender swaps (e.g., converting a bass male voice to a soprano female voice usually requires a +12 shift).
  • Subtle Tuning: If your singer sounds slightly "squeaky," try lowering the pitch by 1 or 2 semitones. If they sound too "breathy," try increasing it.

When to adjust: Always adjust the pitch if the source singer and target voice have different natural ranges. A deep voice trying to hit high notes without a pitch shift often results in "robotic" strain.

2. Fine-Tuning Voice Strength (0.00 to 1.00)

This setting (often called "index rate" in technical circles) determines how much of the target voice's unique characteristics are applied.

  • 0.50 (Natural): A safe, conservative setting. Good for voices that are already similar to the source.
  • 0.75 (Character): Our recommended default. Strong voice identity without losing too much vocal flexibility.
  • 0.90+ (Maximum): Use this if the AI doesn't sound "quite like the person" yet. Be careful: high settings can introduce "metallic" noises or artifacts.

Pro Combo for Realism:

  • For Male to Female: Set Pitch to +12, Voice Strength to 0.70.
  • For Female to Male: Set Pitch to -12, Voice Strength to 0.80.
  • For Same Gender: Keep Pitch at 0, and experiment with Voice Strength between 0.65 and 0.85.

Remember: every voice is a unique fingerprint. Don't be afraid to generate multiple versions with slight variations to find the perfect mix!