Voice

Our goal is to capture your voice as clearly as possible, free from background noise, distractions, or microphone issues. Let this guide your choices—ensure every detail supports clean, uninterrupted audio.

A clear, high-quality voice sample is essential. It will be used for user interactions and to extract content from your audio and video files, ensuring accurate training of your Delphi.

​

Voice Sample Checklist

• Duration:

  • Record at least 90 seconds; 2-4 minutes is ideal.

  • One sample is best.

    • If you use multiple samples, ensure they are recorded with the same microphone, in the same environment, and with identical settings to avoid robotic-sounding audio.

• Microphone:

  • Use a professional microphone for the clearest results.

  • If unavailable, use your iPhone with the Voice Memo app.

• Noise-Free Environment:

  • Record in a space with no background noise. Even minor noise can reduce quality.

• Speaking Style:

  • Speak naturally in the tone you want your Delphi to use.

  • Suggested topics: share your life story, give a motivational speech, or read from a favorite book.

  • Exaggerate emotion slightly to help the AI mimic your style.

​

Adjusting Settings

Stability - This controls the fluctuation of pitch in your voice. If you want to create more variance, move the slider to ‘More Dynamic.’

Clarity- Matches how closely the voice should adhere to the sample(s) when replicating. If you followed the instructions above, we recommend setting this to high.

Style Exaggeration - Determines the cadence of speech. If you would like to have more variance in pace, move the slider to ‘Exaggerated.’

Speaker Boost - We recommend you select this option. It increases the similarity of the voice to the sample.

Voice Description - Add custom descriptions of your voice to put on the finishing touches. Examples include deep, cheerful, raspy, etc.

​

Professional Voice Cloning

Professional voice cloning costs a one-time payment of $6k. It includes a customized voice model trained on 2-3 hours of audio content. Email us at [email protected] if interested.