Voice styles and custom instructions
Updated September 15, 2026
Voice styles change how the chosen voice delivers the text, for example calm, formal or like a news reader. Pick a style from the Style dropdown in the post editor’s AI TTS box, or Select a voice style on the Generator page. Styles only work when Voice Model is set to GPT-4o mini TTS (New); with TTS-1 the option is hidden.
Available style presets
Free: Default, Conversational, Formal, Casual, Whispering.
PRO: Promotional, Advertisement, Motivational, Tutorial, Educational, Inspirational, News Report, Interview, Podcast, Narration, Audiobook, Storytelling, Happy, Sad, Crying, Angry, Excited, Calm, Serious, Comedian, Dramatic.
Each preset is a short written instruction sent to OpenAI along with your text. For example, News Report asks for the voice to “sound like a news reporter”. The model interprets these instructions, so the effect can be subtle or strong depending on the voice and content. Try a couple of styles on the Generator page with a short paragraph before committing to a long post.
Instructions added to every request
With GPT-4o mini TTS the plugin always adds its own base guidance after the style: pause briefly between sentences and paragraphs, avoid long pauses, include emotion rather than a monotone, and speak clearly at a natural pace. The Default style sends only this base guidance.
Write custom voice style instructions (PRO)
The Custom Voice Style Instructions field lets you describe the delivery in your own words. Your text is added after the style preset and the base guidance, so you can combine both. Useful things to ask for include:
- An accent: “Speak with a soft Scottish accent.”
- Pace: “Read slightly slower than normal, as if explaining to a beginner.”
- Tone: “Warm and reassuring, like a customer support agent.”
- Emphasis: “Stress product names clearly.”
Keep instructions short and specific, and avoid instructions that contradict the preset (for example “whisper” with the Excited style). Instructions affect delivery only; they don’t change which words are read. To change the words, use custom pronunciations or the [tts say=""] shortcode.
Set a default style (PRO)
Choose the style that’s pre-selected in AI Text-to-Speech › Settings › Automation › Default Voice Style. This is also the style used when audio is generated automatically on publish. After an administrator generates audio, the style they used becomes the saved default.