Transcription stopped being the hard part years ago; all three are accurate enough on clear audio. The question is what you need the text for: remembering a call, publishing a video in three languages, or editing footage by deleting words. Pick the job, and the tool is obvious.
Side by side
| Tool | The job it is really for | Free tier | Roughly | Deep link |
|---|---|---|---|---|
| Notta | Meetings and calls, with the decisions pulled out | Yes | Free tier, then paid | Visit |
| HappyScribe | Subtitles and translation at volume | Limited | Per minute, or subscription | Visit |
| Descript | Editing video by editing the transcript | Yes | ~$15-30/mo | — |
Pick by what you are actually doing
You keep half-remembering what a client asked for
Notta. It is built around the call rather than the file: it records, transcribes, and pulls out what was actually decided. That last part is the difference between a transcript you never open and notes you use.
Try NottaYou publish video and want subtitles in more than one language
HappyScribe. Transcription, subtitles and translation in one pass, with human linguists available when the output has to be right rather than roughly right. Per-minute pricing means it costs nothing while you are small.
Try HappyScribeYou cut talking-head video or screencasts
Descript. You edit the video by editing the text, and deleting a sentence deletes it from the footage. If that is your workflow it is not really a transcription tool, it is your editor.
Where this comes from
Playbook 03 — The AI Mastery Playbook
Transcription is one input. The book is about what you do with the output — turning repeated work into systems, and verifying what comes back before it reaches a client.
Read it — €20 All eight — €50The Short Version
Every conclusion from all eight playbooks. 97 chapters, 46 pages, free.