At a glance
| First thing to check | Whether the interface you are in accepts audio at all — this changes over time |
|---|---|
| Common blockers | Unsupported file, oversized upload, long recording, codec, account tier, outage |
| Do not assume | That your file is broken — the workflow may simply not be supported right now |
| Reliable path | Use a dedicated transcription tool that is built to accept audio and video |
| Last verified | July 2026 — ChatGPT/OpenAI behavior changes; re-check current docs |
Quick diagnostic
Before blaming your recording, confirm that the exact ChatGPT interface you are using is actually meant to transcribe an uploaded audio file. AI assistants change their supported inputs frequently, and a workflow that worked last month may no longer be available in your plan, app version, or region. Work through the symptom table below, then rule out file, account, and service issues in order.
| Symptom | Likely cause | Fix |
|---|---|---|
| No option to attach or read audio | The current interface may not support audio transcription at all | Verify the feature exists in your plan/version; otherwise use a dedicated tool |
| "Unsupported file type" on upload | Format or container the interface does not accept | Re-export to a common format (MP3/M4A/WAV) or use a transcription tool |
| Upload stalls or fails | File too large for the upload limit | Compress or split the file; check the size cap |
| Only part of a long recording is handled | Duration or token/context limits | Split into shorter segments, or use a tool built for long audio |
| Reply describes the file instead of transcribing | Model is reasoning about, not transcribing, the audio | Confirm a true transcription path is available; do not rely on a chat reply |
| Works sometimes, fails other times | Temporary outage or rate/account limit | Check service status and your usage limits, then retry |
Step 1 — Check the current ChatGPT audio capability (it changes)
Start here, because the most common reason "ChatGPT transcription is not working" is that the specific workflow is not supported in your current setup. Whether an AI chat assistant can take an uploaded audio file and return a transcript depends on the model, the app version, your plan, and the platform — and providers add and remove these abilities without notice.
- Confirm in the current official documentation whether uploaded-audio transcription is offered in your interface today.
- Check whether you are in a text chat, a voice-conversation mode, or a file-analysis mode — these are different features with different inputs.
- Note that a voice mode that lets you talk to the assistant is not the same as transcribing a recording you already have.
- If the capability is not clearly available, stop troubleshooting your file — the workflow itself may simply not exist for you right now.
Step 2 — Check file type, codec, size, and duration
If audio transcription is supported and it still fails, the input is the next suspect. General AI chat interfaces tend to accept a narrower set of formats and smaller files than a purpose-built transcription tool, so a container or codec that plays fine on your device can still be rejected.
- File type: common formats like MP3, M4A, and WAV are the safest; unusual containers or raw codecs are more likely to be refused.
- Codec: a familiar extension can still hold a codec the uploader cannot read — re-encoding often fixes it.
- Size: large files may exceed an upload cap and appear to "hang" rather than error clearly.
- Duration: long recordings can run into length or context limits, so only part of the audio gets handled.
Step 3 — Check account, model, and interface
Account and model differences quietly gate features. The same file can succeed on one plan or model and fail on another, and a mobile app can differ from the web or desktop experience.
- Plan/tier: some capabilities and higher limits are reserved for paid tiers.
- Model: not every model exposes the same file-handling or transcription behavior.
- Interface: web, desktop app, and mobile app versions can support different inputs — update the app before assuming a bug.
- Usage limits: you may have hit a message, upload, or rate cap that will reset later.
Convert or re-export the file safely
When the format is the blocker, re-export from the original source rather than converting an already-compressed copy. Fewer generations of re-compression means better recognition later.
- Export from the original recorder or editor to MP3, M4A, or WAV.
- Keep the source file — do not delete the original in case you need a cleaner export.
- Avoid stacking conversions (e.g. a shared, re-compressed chat copy of a chat copy).
- If you only need to check quality, a short test export confirms the format is accepted before you process the whole thing.
A dedicated transcription tool typically decodes a wide range of inputs for you — MP3, WAV, M4A, AAC, FLAC, OGG, Opus, WebM, and the audio track of video files — so re-exporting is often unnecessary in the first place.
Split a long recording without losing context
If a long file only partly transcribes, the cause is usually a length or context limit rather than a corrupt file. Splitting the recording into shorter segments works, but do it so you do not lose the join between parts.
- Cut at natural pauses or speaker changes, not mid-sentence, so nothing is clipped.
- Add a few seconds of overlap between segments so words at the boundary appear in both.
- Label segments in order so you can reassemble the transcript reliably.
- For anything more than a few minutes, prefer a tool built to handle long audio in one pass.
Check service status and browser issues
Intermittent failures that come and go are often outside your file entirely. Rule out the service and your browser before more re-exporting.
- Check the provider status page for a current incident or degraded uploads.
- Retry after a short wait — a transient error may clear on its own.
- Try a different browser or an incognito window to rule out extensions and cached state.
- Confirm a stable connection; a dropped upload on a weak network can look like a file error.
Privacy and confidential audio
Before uploading sensitive audio anywhere, consider where it goes and how it is retained. Different providers have different data-handling and training policies, and those terms change — do not assume a general AI chat tool treats a confidential recording the way a dedicated transcription service does.
- Read the current retention and data-use terms of whatever tool you upload to.
- Recording other people may require their consent depending on your jurisdiction.
- Prefer a tool that states its retention and training terms plainly and lets you delete files.
When to use a dedicated tool
If you have confirmed the workflow is supported, fixed the format, and still hit walls, a purpose-built transcription tool is the pragmatic answer. It is designed to accept audio and video, handle long recordings, and return an editable transcript with timestamps — the exact job a general chat assistant may not be built to do.
| Need | General AI chat | Dedicated transcription tool |
|---|---|---|
| Purpose-built for audio in | Not guaranteed; varies by version | Yes — audio and video are the input |
| Wide format support | Often narrow | MP3, WAV, M4A, AAC, FLAC, OGG, video audio, and more |
| Long recordings | Length/context limits common | Built to process long files |
| Editable transcript with timestamps | Inconsistent | Word-level timestamps on uploads/recordings |
| Exports (TXT, DOCX, PDF, SRT, VTT) | Usually not | Yes (full export is a paid feature) |
You can test this on your own file right now: drop it into the uploader at the top of this page. Without an account, the first 5 minutes of a file up to 50 MB are transcribed so you can confirm it works before committing to anything.
Frequently Asked Questions
Why can ChatGPT not transcribe my audio?
The usual reasons are that the interface you are using does not support uploaded-audio transcription right now, the file type or size is not accepted, the recording is too long, or there is a temporary outage or account limit. Confirm the feature is available before assuming your file is at fault. Behavior changes over time, so verify against the current product.
ChatGPT can't transcribe audio at all — is that normal?
It can be. Whether an AI chat assistant accepts an audio file for transcription depends on the model, plan, app version, and region, and providers add or remove this ability without notice. If there is no clear option to transcribe an uploaded recording, the workflow may simply not be supported for you, not broken.
I keep getting a ChatGPT transcription error — how do I fix it?
Work in order: confirm audio transcription is supported in your interface, re-export the file to a common format like MP3 or M4A, check size and length limits, verify your plan and app version, and rule out a service outage. If it still fails, use a dedicated transcription tool built to accept audio and video.
Why is my ChatGPT audio upload not working?
Uploads commonly fail because the format or codec is not accepted, the file exceeds the size cap, the connection dropped, or the interface does not support audio uploads in the first place. Try a smaller, common-format test export, check the current limits, and switch browser or network to isolate the cause.
Does ChatGPT support long audio recordings?
Long recordings often run into length or context limits, so only part of the audio is handled even when the upload succeeds. Splitting the file into shorter segments with a little overlap can help, but a tool built to process long audio in one pass is more reliable for anything beyond a few minutes.
Which audio formats should I use to avoid errors?
Common formats like MP3, M4A, and WAV are the safest across tools. Unusual containers or raw codecs are more likely to be rejected even if they play fine on your device. Re-export from the original source rather than converting an already-compressed copy.
Is it safe to upload confidential audio to an AI chat tool?
Check the current retention and data-use terms of whatever tool you use, since policies differ and change. For sensitive recordings, prefer a service that states its retention and training terms plainly and lets you delete files, and confirm consent rules for recording others in your jurisdiction.
What is the most reliable way to transcribe a file?
Use a purpose-built transcription tool. It is designed to accept audio and video, handle long recordings, and return an editable transcript with timestamps and exports (TXT, DOCX, PDF, SRT, VTT). You can test your own file free for the first 5 minutes without an account.
Related resources
Reviewed by the TranscribeThis product team · Last updated: July 2026