Spokenly Logo
Spokenly
ChatGPT Guide

Can ChatGPT Transcribe Audio? How to Get a Transcript

Get a transcript on any ChatGPT plan: upload the file on a paid plan, or transcribe it with Spokenly first.

Download Spokenly

Method 1: Spokenly Plugin in ChatGPT

Whether you can install plugins depends on your ChatGPT plan, workspace, and region. The Spokenly plugin opens a transcript view inside the chat, and ChatGPT reads the whole transcript to answer your questions. The plugin needs no Spokenly account or API key.

  1. 1In ChatGPT, open Plugins in the sidebar, find Spokenly, choose Install plugin, and connect the plugin.
  2. 2Type @Spokenly or pick Spokenly from the + menu. Ask it to open the transcriber and choose an audio file. On a paid plan, you can also attach the recording to your message.
  3. 3When the transcript appears, click a line or timestamp to listen from that point.
  4. 4Ask ChatGPT for a summary, action items, or exact quotes from the recording.

The plugin accepts MP3, WAV, M4A, MP4, WebM, OGG, and FLAC files. Groq transcribes the audio, with fal.ai as the fallback. Spokenly deletes the audio and transcript from its servers within an hour, and your ChatGPT data controls apply to the copy in your chat. Setup details are on the Spokenly for ChatGPT page.

Method 2: Transcribe in the Spokenly App, Then Ask ChatGPT

This works on every ChatGPT plan, including Free. The Spokenly app transcribes audio and video files on Mac, Windows, Linux, and iPhone. With a local Parakeet or Whisper model, transcription is free, works offline, and keeps the recording on your device. That suits long or confidential recordings.

  1. 1Download Spokenly and choose a local model, or a cloud model with your own API key.
  2. 2On a computer, open Transcribe File and add the recording. On iPhone, share the file to Spokenly.
  3. 3Copy the transcript. On a computer, you can also export it as TXT, SRT, or VTT.
  4. 4Paste the text into ChatGPT and ask for a summary, notes, or a draft.

For more ways to transcribe a recording, see Audio to Text.

Method 3: Upload the Audio File to ChatGPT (Paid Plans)

  1. 1In a new chat, attach the audio file to your message.
  2. 2Ask ChatGPT to transcribe it word for word. Add "with speaker labels" if you need them.
  3. 3Check names, numbers, and technical terms against the recording before you rely on the text: transcripts can contain errors, and speaker identification may be unreliable.

Formats

WAV, MP3, M4A, AAC, FLAC, OGG, and PCM, plus WebM and MP4 files that contain only audio.

Size

Up to 512 MB per file.

Method 4: Record a Meeting on Mac

ChatGPT Record captures your microphone and system audio in the ChatGPT app for macOS, transcribes the audio live, and saves a summary to your chat history. Plus, Pro, Business, Enterprise, and Edu plans include it, and Enterprise and Edu admins must turn it on. A session runs up to four hours, and ChatGPT labels the speakers and deletes the audio after transcription.

Dictation and Voice Mode Do Not Transcribe Files

The microphone in ChatGPT's message box turns your speech into editable text for your next prompt. Voice mode is a spoken conversation that leaves a rough transcript in the chat. To dictate into ChatGPT and other apps with a speech model you choose, see speech to text for ChatGPT.

Why ChatGPT Says It Can't Transcribe Audio

You are on the Free plan

Use Method 2, or Method 1 if your plan allows plugins.

You see no upload option

Availability still varies by app version, region, workspace, and model. Update the app or try chatgpt.com in a browser.

ChatGPT rejects your MP4 or WebM file

ChatGPT treats a file with a video track as video. Export only the audio as an M4A or MP3 file, or use Method 1 or 2, which accept video files.

ChatGPT cannot read your file

The file may be damaged or use an unsupported codec. Convert it to MP3 or WAV and upload it again.

A long recording times out

Split the file into shorter parts, or transcribe it in the Spokenly app.

For Developers: The OpenAI Speech to Text API

In the OpenAI API, the recommended model for recorded speech is gpt-transcribe. It costs $0.0045 per minute and accepts files up to 25 MB. The whisper-1 model and the GPT-4o transcription models will stop working on February 26, 2027. For model details, see the GPT-Transcribe guide.

Can Claude or Gemini Transcribe Audio?

Claude

Claude accepts documents and images, not audio. The Spokenly connector in Claude's directory adds a transcript view like the one in the ChatGPT plugin (Spokenly for Claude).

Gemini

The Gemini app accepts audio uploads: up to 10 minutes of audio per prompt, or 3 hours with Google AI Pro or Ultra.

Sources: ChatGPT release notes, OpenAI Help Center: Plugins in ChatGPT, OpenAI API pricing, Claude Help Center: Upload files. Checked October 19, 2026.

Ready to try Spokenly?

Free to use with local models. No account required.

Download Spokenly
Mac, Windows, Linux & iPhone
Free local models
Works offline

Read next