Voiced Conversation – OpenAI + ElevenLabs

This sequence generates a two-character scripted conversation using OpenAI and ElevenLabs for text-to-speech conversion. It combines human input for customization and AI for generating content and voice synthesis. Below is a step-by-step explanation of the workflow.

1. Defining the Conversation Topic

• Step: The sequence begins by asking the user:
  • “What is the conversation about?”
• User Input: The user provides the topic of the conversation, which defines the context for the generated dialogue.
• Purpose: This ensures the AI-generated script is aligned with the user’s goals.

2. Generating the Scripted Dialogue

• Process: The sequence uses OpenAI to generate a short dialogue between two characters based on the provided topic.
• Prompt Details: The AI is instructed to:
  • Script a conversation between two characters based on the provided topic.
  • Limit each line to a maximum of 10 words.
  • Alternate lines between characters.
  • Use — to indicate pauses for natural flow.
  • Produce a conversation with exactly 6 lines.


3. Human Review and Editing

• Step: The user is presented with the generated script and asked:
  • “Please edit as needed.”
• Functionality:
  • Users can modify the script to refine tone, content, or phrasing.
  • The original script is also provided for reference under Review Material.

4. Text-to-Speech Conversion

• Step: The finalized script is converted into audio using ElevenLabs text-to-speech technology.
• Voices Assigned:
  • Two unique voices (predefined in the sequence) are assigned to the characters:
    • Voice 1: For odd-numbered lines (e.g., lines 1, 3, 5).
    • Voice 2: For even-numbered lines (e.g., lines 2, 4, 6).
• Process: The script lines are split and matched with the respective voices. Each line is processed individually to produce high-quality audio.

5. Iterative Dialogue Playback

• Step: The sequence loops through each line of the script, applying text-to-speech processing for all six lines.
• Logic:
  • A mathematical node determines which voice to use based on whether the line number is odd or even.
  • This ensures the correct alternation of voices.

6. Sequence Completion

• Step: After processing all six lines of dialogue, the sequence ends.