Should I change every audio setting before testing again?

Change one thing at a time and repeat a short, harmless request. That gives you a clearer record of which adjustment helped.

What details make a useful voice troubleshooting note?

Keep the device, browser or app, the result you expected, the exact symptom and any error message. Add the checks you tried and their outcomes, without including private conversation content.

Should I save the entire conversation before another test?

A checked summary is usually more useful for continuing your work. Keep the goal, accepted conclusion, open question and next action, and verify important wording yourself.

Will a fenbs connection restore spoken replies?

No. fenbs keeps useful work available to connected assistants. Microphone and audio problems still need to be resolved in the device, browser or ChatGPT setup.

ChatGPT voice not working? Check access and sound

Start with the Voice entry point, microphone permission and audio checks. Then separate interruptions and account limits from a device problem, and keep the useful work before trying again.

June 27, 2026Updated October 28, 20264 min read

If ChatGPT voice is not working, check that you opened Voice, allowed microphone access and can hear your device audio. Then check whether the conversation ended because of a limit rather than a microphone fault. Work through one symptom at a time using OpenAI’s Voice guide (opens in a new tab). Tools like fenbs keep the useful work outside the voice session, so you can continue later without explaining everything again.

What should you check first?

On ChatGPT.com, select the Voice icon in the prompt window. On iOS or Android, select the Voice icon in the message bar.
Allow browser or app microphone access when prompted.
Try one short sentence and listen for the answer. Note whether the problem is hearing you, playing audio or keeping the conversation going.
Change one thing at a time, then repeat the same short test. Keep a note of what changed and what happened.

Use a harmless test such as asking for three names for a fictional garden. Avoid restarting a long explanation of your actual work each time. Once the test works, our guide to talking to ChatGPT covers making a working conversation useful.

Are you using Voice or recording a text prompt?

OpenAI distinguishes Voice from Dictation (opens in a new tab): Voice is a spoken conversation; Dictation creates editable prompt text from a recording. Text appearing after a recording does not, by itself, mean spoken replies have failed.

Before changing anything, write down the result you expected. “I want to speak a draft and edit it before sending” is different from “I want to hear an answer.” That small distinction helps you test the right workflow instead of trying to repair a feature you did not open.

For troubleshooting, use the same simple request throughout. Changing both the request and the audio setup makes it harder to tell which change helped. Treat a successful short exchange as a checkpoint before returning to a longer discussion.

What if ChatGPT cannot hear you?

Start with the microphone permission from the first check. Confirm whether you granted access when prompted and whether you are testing the device you meant to use. Keep a refused permission separate from a conversation that starts but fails to understand your words. Those are different observations to include in your record.

As a diagnostic check, identify the microphone you intend to use, especially when headphones or other audio equipment are connected. Note whether the failure happens with that equipment connected, without assuming a particular accessory is responsible. This is a way to describe the problem precisely, not a guaranteed cure.

Record the device, browser or app, expected input and observed result. “Nothing happens after I speak” is a more useful starting point than “voice is broken.” If recognition works but the words are wrong, keep that as a separate symptom rather than treating it as complete microphone failure.

What if replies are silent or keep getting interrupted?

For interruptions, OpenAI recommends headphones, a quieter setting and checking device volume (opens in a new tab). On iPhone, it also documents Control Center > Mic Mode > Voice Isolation. Settings > Voice > Language sets your preferred language.

Try those checks separately. First note whether you hear nothing at all or only part of a reply. Then test a short exchange in a quiet place. If you change headphones, volume and surroundings at once, a successful result will not tell you which adjustment mattered.

Keep the test modest: ask one question, wait and record the outcome. Do not use an important interview or an unrepeated spoken instruction as your audio test. If the answer is needed for your work, check its substance separately from whether it sounded clear.

Why did the call end or a Voice option disappear?

Only one Voice conversation can run at a time. Usage limits, maximum session lengths or limits on how much a conversation can hold can end a call. Available options are selected under Settings > Voice and depend on account conditions; Live does not support screen sharing. Check OpenAI’s current options and limits (opens in a new tab) before diagnosing missing controls as a device fault.

A linked teen account can also have Voice Mode managed through parental controls (opens in a new tab). Our ChatGPT parental-controls guide covers that separate setup. Do not work around a family or workplace restriction while troubleshooting audio.

Copy any limit or restriction message into your own troubleshooting record, leaving out private details. If you notice you are using a different account, use the account-switching guide rather than mixing account changes into the microphone test.

What should you keep before trying again?

Voice transcripts are not verbatim records (opens in a new tab). Review any useful wording before relying on it. Our advice is to save a short checked summary: the goal, the conclusion you accepted, the question still open and the next action.

What you were trying to finish, such as outlining a presentation.
What you checked and the exact symptom that remains.
What you want to continue, separate from the audio problem.
Which suggested fix you actually tested and whether it helped.

A startup founder might keep the approved outline and the next drafting task; our ChatGPT for startups guide explores that work. For a different device-centered workflow, see ChatGPT versus Apple Intelligence. Keeping the checked work in fenbs lets the next assistant read it without replaying the conversation.

How fenbs helps

Never lose the work you did in an AI session

After a useful AI session, save it in fenbs: the key observations, findings, research and context. Then carry on from exactly where you left off, on any device and in any other AI. The next AI you connect reads it at the start of the session, so you never have to explain the same problem again.

Create a free account

Free for up to 5 people and 50 open tasks · No credit card · Works with Claude, ChatGPT, Cursor and more

How to keep voice-session notes with fenbs?

Keep what you need to carry on from exactly where you left off, on any device and in any other AI. Save the useful work and the next step, not a pile of unreviewed troubleshooting guesses.

Create a free account.
Write down what you finished, what still needs checking and what to do next.
Connect ChatGPT to fenbs, then ask it to read where you left off before continuing.
Before stopping, ask your connected assistant to save what changed and the next step. Check that the summary matches what you meant.

fenbs does not repair microphones or import ChatGPT chats. If you only need an audio setting corrected, keep working through the checks above. It helps when a useful conversation has produced work you want another session to continue. Find more practical workflows in AI at work.