How to transcribe audio for free
A practical guide to turning MP3, M4A, WAV, WebM, and video audio into a clean transcript without building a transcription workflow yourself.
Audio transcription is useful only when the output is easy to review, export, and reuse. A raw text block is rarely enough. For real work, you need the source audio, a timestamped transcript, and a predictable way to export or share the result.
The fastest path
- Open Orpheus and upload an audio or video file.
- Use the free tool for short tests, or sign in when you need job history and exports.
- Wait for the transcript to finish.
- Review the timestamped segments.
- Export the result as text, JSON, VTT, or SRT depending on your workflow.
For long recordings, the production path is asynchronous: upload the file, create a job, and receive the result by status polling or webhook. That avoids browser timeouts and makes large files recoverable.
Which files work best?
MP3, M4A, WAV, WebM, OGG, and FLAC are the safest formats. Clean speech, moderate volume, and limited background noise improve accuracy. If your recording includes long silence, music, or several people speaking at once, review the transcript before publishing it.
What to check before publishing
- Names, company names, and technical terms.
- Quotes that will be used in public material.
- Captions for legal, medical, education, or accessibility use.
- Segments with overlapping speakers or background noise.
AI transcription is fast, but important content still deserves human review.
When to use the API
Use the public API when transcription is part of your product or internal workflow. Orpheus uses an async-first public API: create a job, process the file reliably, and receive completion through webhook or polling. This is the stable contract for production integrations.