💡 Feature Request: Smart Prompt Optimization Layer for Voice Input

[HongjieYu]

Hi OpenAI team and fellow users

I’d like to suggest a feature that I believe could significantly improve the voice input experience in ChatGPT and similar LLM-based tools.

---

The Idea: Smart Prompt Optimization Layer for Voice Input

When we type, we naturally organize our thoughts, edit on the fly, and refine our language to make prompts clearer. But when we speak, it’s often more casual, fragmented, or spontaneous—which can lead to less precise prompts and, sometimes, less accurate results from the model.

So here’s the proposal:

Add a smart LLM-powered layer between the raw voice input and the final prompt, which restructures, clarifies, and optimizes the spoken content before it’s sent to the model.

---

How it could work:

• Voice is transcribed as usual.
• Before sending the input to the model, it passes through an LLM that rewrites the message into a clearer, more structured prompt.
• UI shows both:
  • “What you said” (original transcript)
  • “What will be asked” (LLM-refined prompt)
• User can edit the optimized prompt or toggle this feature on/off.

---

Benefits:

• Improves the output quality when using voice.
• Makes prompt crafting easier for non-technical or casual users.
• Brings the best of both worlds: the speed of voice + the precision of a well-written prompt.

---

I’d love to hear your thoughts—and if the OpenAI team sees this, I hope it sparks some inspiration! Thanks for all the amazing work you’re doing

— A ChatGPT user and fan ，Hongjie Yu