Can ChatGPT or Gemini transcribe an audio file?

What each route states

1. Gemini app
   Limit it states: Total audio length up to 10 minutes; 3 hours with Google AI Pro or Ultra
   Account or key: Signed in to Gemini Apps
   Subtitle file (SRT/VTT): Not mentioned; the page does not mention transcripts either
2. Gemini 3.5 Transcribe (developer API)
   Limit it states: Files up to 1 hour per request (30 minutes with speaker labels or word timestamps); free of charge on the API free tier per its pricing page
   Account or key: A Google developer API key
   Subtitle file (SRT/VTT): Not mentioned
3. ChatGPT app
   Limit it states: Not stated: OpenAI's help pages returned HTTP 403 to us. Check OpenAI's own pages
   Account or key: Not stated
   Subtitle file (SRT/VTT): Not stated
4. OpenAI speech-to-text API
   Limit it states: 25 MB per file; mp3, mp4, mpeg, mpga, m4a, wav, webm; gpt-transcribe $0.0045 a minute
   Account or key: An OpenAI API account
   Subtitle file (SRT/VTT): Timestamps only on an older model that the page says shuts down on 2027-02-26
5. Word (Transcribe)
   Limit it states: 300 minutes of uploaded audio a month; .wav, .mp4, .m4a, .mp3
   Account or key: Microsoft 365 subscription
   Subtitle file (SRT/VTT): Not stated; the transcript goes into a Word document
6. TryScriptorium
   Limit it states: One file up to 15 minutes and 25 MB, 3 tries a day
   Account or key: No account for a first file
   Subtitle file (SRT/VTT): Yes: SRT and VTT, with TXT, timestamped TXT, SRT, VTT, Markdown, JSON and CSV

Read on 2026-10-27 from each product's own pages (sources below). "Not stated" means the page did not say or could not be read; we did not guess. Google's own pages disagree about Gemini 3.5 Transcribe (its blog says up to three speakers and a public preview, its developer docs say up to eight and its models page says stable), so check Google's pages.

How to choose

• ·A short clip you just want to read: a chat app is fine if its limit covers it; check the limit first.
• ·A developer building a pipeline: the Gemini or OpenAI APIs; compare the file limits above (25 MB for OpenAI's, longer for Google's).
• ·You pay for Microsoft 365 already: Word's Transcribe, within 300 uploaded minutes a month.
• ·A subtitle file, a longer file, or a transcript you can edit line by line: a transcription tool. We make TryScriptorium; Gemini compared and ChatGPT compared say where each is the better choice.
• ·Arabic: Gemini 3.5 Transcribe's docs list Arabic (Egypt) only; see Arabic transcription tools, ranked.

Steps

1. Pick the route that fits the file. A short clip in a chat app, a developer API, Word, or a transcription tool; the table shows what each one states.
2. Check the limit before you upload. Length and file size limits are the usual reason an upload fails; they are in the table.
3. Upload and read the result. For a transcription tool, drop the file on audio to text; one file up to 15 minutes and 25 MB, 3 tries a day, no account.
4. Check names and numbers. Every route here is automatic. Read the text against the audio before you rely on it.

Sources

Every row about another product was read on that product's own page on 2026-10-27. Prices and limits change; confirm on their site before you rely on them. We make TryScriptorium, so read our claims with that in mind.

• https://support.google.com/gemini/answer/14903178
• https://ai.google.dev/gemini-api/docs/transcribe
• https://developers.openai.com/api/docs/guides/speech-to-text
• https://support.microsoft.com/en-us/office/transcribe-your-recordings-7fc2efec-245e-45f0-b053-2a97531ecf57