Voice Direction tells a voice how to perform a specific passage. Write it like a director speaking to an actor: describe what is happening, what the speaker wants, and the vocal choices that make the moment work.
With BreezeBlue, Voice Direction can be written in natural language. It can be short. Two or three useful decisions are usually enough.
Start with the transcript
Read the words before writing the direction. The same sentence can express excitement, fear, sarcasm, awkwardness, or grief. Voice Direction tells the model which interpretation to perform and which audible choices should communicate it.
The direction can still be short. Choose only the details that change the meaning or delivery of this passage.
A compact Voice Direction structure
When you want more control, build the direction from five possible dimensions.
1. Situation
Give the speaker a playable circumstance. They may be hiding from someone nearby, reporting from a dangerous scene, comforting a worried friend, or giving instructions under time pressure. Include only the context that changes how the line should sound.
2. Intent
Describe what the speaker is trying to do to the listener: reassure, persuade, warn, command, conceal, comfort, or provoke. The same calm voice can sound very different when it is reassuring someone instead of hiding a threat.
3. Emotion and intensity
Name the emotional state and give it a boundary. Directions such as restrained panic, mild impatience, or urgent but not panicked are more specific than simply asking for more emotion.
4. Rhythm and emphasis
Describe the pace, pauses, breath, and words that should carry extra weight. For example: Keep it fast and precise, Pause after the warning, or Emphasize “right now” without shouting.
5. Physical and vocal state
The body affects the voice. The speaker may be exhausted, trembling, breathless, holding back tears, or suppressing laughter. Describe the state and the audible effect it should have on the line.
Combine only the dimensions the scene needs
You rarely need all five dimensions in one instruction. Choose the two or three that carry the scene, then combine them in natural language:
Play this as [situation].
The speaker is trying to [intent].
Keep the emotion [state and intensity].
Use [pace, pause, breath, volume, or emphasis].
The speaker sounds physically [state], with [audible effect].Use only the lines the passage needs. Stable characteristics such as age, accent, and core vocal texture belong in voice selection or Voice Design; Voice Direction shapes this particular performance.
Five directions for the same line
The five examples below use the same transcript. Each Direction selects a different combination of the structure above, producing a different interpretation without changing the words.
Transcript
I can’t believe you brought him here!
Whispered and afraid of being overheard
Direction
Say it whispering, very quiet and tense, as if you are afraid of being overheard.
Direction dimensions: Situation · Emotion and intensity · Rhythm and emphasis · Physical and vocal state
Listen: whispered
Excited surprise
Direction
Say it excitedly, with bright surprise and rising energy, as if his arrival is wonderful and completely unexpected.
Direction dimensions: Situation · Emotion and intensity · Rhythm and emphasis
Listen: excited surprise
Sarcastic disbelief
Direction
Say it sarcastically, with dry disbelief and a dismissive tone, as if this was obviously a terrible idea.
Direction dimensions: Intent · Emotion and intensity · Rhythm and emphasis
Listen: sarcastic disbelief
Nervous humor
Direction
Say the line with a nervous laugh after the first phrase, as if trying to hide panic behind humor.
Direction dimensions: Intent · Emotion and intensity · Rhythm and emphasis · Physical and vocal state
Listen: nervous humor
Restrained grief
Direction
Say it while holding back tears, with a cracked voice and emotional pauses, as if bringing him here deeply hurt you.
Direction dimensions: Situation · Emotion and intensity · Rhythm and emphasis · Physical and vocal state
Listen: restrained grief
Choose a Guidance setting
Guidance controls how strongly the generation interprets Voice Direction.
API
guidance_scale supports integer values from 1 to 10. For Voice Direction, 1 to 3 is a useful range for balancing expression and stability. Set the value explicitly when comparing different instructions.
Start lower for restrained delivery. Increase gradually when the result is too neutral, and reduce it when the performance becomes exaggerated.
Web
The web product provides four levels: Stable, Expressive, Creative, and Intense.
Use Stable for controlled, consistent delivery. Use Creative when you want more interpretation or emotional development. Stable and Creative are the recommended starting choices for most generations; use Expressive or Intense when the scene specifically calls for them.