Feature Request: Full Text and Code Support in Voice Mode

ChatGPT Feature requests
,
1

Feature Request: Full Text and Code Support in Voice Mode

The Problem

Voice mode is excellent for collaborative problem-solving, but it becomes frustrating when a conversation produces something that needs to be used outside the conversation, such as a complete code file, a technical specification, or a feature request.

I recently encountered this while working with ChatGPT on a custom Hubitat home automation project. I wanted to discuss the design conversationally, have ChatGPT develop a complete Groovy application, and then provide feedback on the implementation.

The conversation worked well until I needed the actual written output. Getting a substantial block of code or text into a usable form required switching out of voice mode and continuing in text.

The irony is that I then asked ChatGPT to help me submit a feature request about this exact limitation. Even though we had discussed the request in voice mode, I had to switch to text mode to get the final request written out properly.

I had to leave voice mode to explain why I shouldn’t have to leave voice mode.

The Proposed Solution

Allow users to access, view, and copy substantial text outputs directly within the voice-mode interface.

Ideally, voice mode would support a persistent text panel alongside the conversation, allowing users to:

  • View complete code files, long-form documents, and other substantial text outputs.

  • Copy the full output without truncation or cumbersome workarounds.

  • Continue speaking with ChatGPT while reviewing, editing, and refining the written output.

  • Move seamlessly between conversational collaboration and practical work without restarting or abandoning the voice session.

This would be particularly valuable for programming, technical troubleshooting, writing, research, and other tasks that combine spoken collaboration with substantial written deliverables.

Why It Matters

Voice mode shouldn’t be limited to conversations that end with a spoken answer. It should also support conversations that produce something tangible.

One of ChatGPT’s greatest strengths is its ability to work collaboratively through complex problems over multiple iterations. Voice makes that collaboration feel natural. But when the result is code, a document, or another practical deliverable, the interface should make that result just as accessible as it would be in text mode.

The goal isn’t simply to make voice mode talk better. It’s to make voice mode a complete working environment, where users can discuss, create, inspect, and use the results without breaking the flow of collaboration.

2

Hi! I’m very interested in being able to continue speaking while the material remains readable and usable.

In my work with Cristallo, my AI assistant, I review texts and other materials through an ongoing dialogue. I may ask him to revisit a sentence, clarify a passage, or compare a formulation with the source it comes from. The revision takes shape as we discuss it: sometimes we change the text; at other times, we realise that we need to explore the content further first.

For example, when reviewing an article, a sentence may be too broad compared with what a source actually documents. Cristallo and I compare the passage, consider more precise wording, and reread the paragraph to see whether the change preserves the meaning of the text.

In a voice interface, it would be useful to follow this process with the document open alongside the conversation: the full text remains available to consult, the passage under discussion is highlighted, and the proposed wording can be compared with the current version.

It should be clear which part of the material is being discussed. If someone says “let’s revise this sentence,” it would be useful to see which sentence has been identified and be able to correct that reference immediately while continuing to speak.

The proposal and the adopted change should also remain distinct: discussing a formulation does not mean it has been approved for inclusion. At the end of the discussion, it should be possible to copy or save the complete result without losing the agreed revisions.

For me, this would be a practical way of making the dialogue and the material part of the same work. In your coding example, a similar view could keep the complete file accessible while highlighting the function or block being discussed.

Would being able to consult the entire output and identify the passage currently being worked on address your needs too?