Can Claude transcribe audio or video?
No, Claude cannot transcribe audio or video files on its own. ChatGPT and Gemini can, or you can add a transcription connector to Claude.
No, Claude can't transcribe audio or video: its uploads accept documents and images only, not sound files. To transcribe, use ChatGPT on a paid plan or Gemini, or add a connector such as ElevenLabs to Claude, then let Claude clean up and summarize the text.
Good for
- Podcasters and creators who need transcripts of recordings
- Students and researchers transcribing interviews and lectures
- Teams that want meeting transcripts searchable in Claude
- Anyone turning a video into notes or a blog post
The short answer
To transcribe, use ChatGPT on a paid plan for long files or Gemini for short free ones, then bring the text to Claude if you prefer it for summaries. Claude users who want it in one place can add the ElevenLabs connector.
Our pick: ElevenLabs
Claude does not take audio, so the honest pick for transcription is ChatGPT (paid) or Gemini (free up to 10 minutes). The main limit is length and accuracy: long files can time out, and results vary by language.
Best for
- Interview and podcast transcripts
- Meeting notes from Granola recordings
- Turning YouTube videos into notes
Skip it if
- You expect Claude to accept an MP3 directly
- You need a free tool for recordings over 10 minutes
- You need certified or legal-grade transcripts
Get transcripts into Claude with a connector
Claude
- 1
Add ElevenLabs for audio files
In Claude, click +, hover Connectors, choose Manage connectors, add ElevenLabs and sign in. It can turn an audio file into text and spends your ElevenLabs credits.
Uses ElevenLabs
- 2
Or add Granola for meetings
If you record meetings with Granola, add its connector the same way. Claude can then search your meeting notes and, on paid Granola plans, quote the transcript.
PromptFind last Tuesday's call with Acme in Granola and list every action item with who owns it.Uses Granola
- 3
For YouTube videos, pull the transcript
In Claude Desktop or Claude Code, a YouTube transcript server fetches a video's captions with no API key. Install it from its page.
PromptGet the transcript of this YouTube video and turn it into a 600-word blog post with headings: [link] - 4
Or paste a transcript you already have
Attach a TXT, DOCX or PDF transcript, or paste the text, and ask Claude to clean it up.
PromptClean up this transcript: fix punctuation, remove filler words, keep the speaker labels, and add section headings every topic change.
Transcribe audio or video free in Gemini
Gemini
- 1
Upload the file
At gemini.google.com, click Add Files and choose your audio or video. Video files can be up to 2GB and other files up to 100MB.
- 2
Mind the length
Free accounts can upload up to 10 minutes of audio or 5 minutes of video in total. Google AI Pro or Ultra raises it to 3 hours of audio and 1 hour of video.
- 3
Ask for a transcript
Ask for the exact format so you do not get a summary instead.
PromptGive me a full verbatim transcript of this recording with speaker labels and timestamps. Do not summarize.
Transcribe an audio file in ChatGPT
ChatGPT
- 1
Check your plan and file
Audio uploads need a paid ChatGPT plan. Use MP3, M4A, WAV, FLAC, AAC, OGG or audio-only MP4, up to 512MB. If you have a video, export the audio first.
- 2
Attach the recording
Select + in the prompt area, then Add photos & files, and choose the audio file.
- 3
Ask for the transcript format you need
Say whether you want speaker labels, timestamps or a clean read.
PromptTranscribe this interview word for word. Label speakers as Host and Guest, add a timestamp every 2 minutes, and mark anything you could not hear as [inaudible]. - 4
Then work with it
Ask follow-up questions about the recording in the same chat.
PromptNow give me a 5-bullet summary and the 3 best quotes for social media, with timestamps.Tip: Very long recordings may time out. If it fails, split the file into shorter parts.
Best plugs for this
Plugs we hold that produce transcripts, ranked by official status and reviewed page first, then GitHub stars.
- 1
ElevenLabsOfficialHostedOAuthElevenLabs' own hosted connector: turns an audio file into text inside Claude or ChatGPT. Uses your ElevenLabs credits.
- 2GranolaOfficialHostedOAuth
Granola's own connector: search meeting notes and quote transcripts from Claude or ChatGPT. Transcripts need a paid Granola plan.
- 3
mcp-youtube-transcriptLocalFetches YouTube transcripts with no key, so Claude can summarize or repurpose a video.
494 GitHub stars - 4
mlx-whisper-mcpRuns Whisper transcription on your own Mac with Apple silicon, so audio never leaves your computer.
24 GitHub stars - 5
Video TranscriberLocalTranscribes local files and videos from many sites offline, and searches across your transcripts.
19 GitHub stars
Questions
Can Claude transcribe audio?
No. Claude accepts documents and images, not audio files. Add the ElevenLabs connector to transcribe inside Claude, or transcribe in ChatGPT or Gemini and give Claude the text.
Can Claude transcribe a video?
No. Claude does not accept video uploads. For YouTube, a transcript server can fetch the captions for Claude; for your own videos, transcribe them in Gemini or ChatGPT first.
Which AI can transcribe audio?
ChatGPT transcribes audio files on paid plans, up to 512MB per file. Gemini transcribes audio and video uploads, up to 10 minutes of audio free or 3 hours with Google AI Pro or Ultra.
Can ChatGPT transcribe audio for free?
No. Audio uploads need a paid ChatGPT plan; Free accounts cannot upload audio files.
Is there a free AI that transcribes audio?
Gemini's free tier transcribes up to 10 minutes of audio per upload. For longer files without paying, a local Whisper server runs on your own computer.
What audio formats does ChatGPT accept?
WAV, MP3, OGG, FLAC, AAC, M4A, PCM, audio-only WebM and audio-only MP4. MP4 or WebM files with video are rejected, so export the audio track.
How do I transcribe a long recording?
Split it into shorter parts. ChatGPT may time out on very long recordings, and Gemini's limit is 3 hours of audio with a paid Google AI plan.
Can Claude summarize a transcript?
Yes. Paste the transcript or attach it as a TXT, DOCX or PDF file, and Claude can summarize it, pull quotes or turn it into notes.
How accurate is AI transcription?
It varies with audio quality and language. OpenAI notes that transcription performance varies across languages, so check names, numbers and quotes against the recording.
Sources
Docs, issue trackers and discussions this page is built from.
Show all 9Hide
- Upload files to Claude · support.claude.com
- Uploading files and audio to ChatGPT · help.openai.com
- Upload & analyze files in Gemini Apps · support.google.com
- ElevenLabs creative studio tools · github.com
- Granola MCP · docs.granola.ai
- Use connectors to extend Claude's capabilities · support.claude.com
- GitHub · github.com
- GitHub · github.com
- GitHub · github.com