Transcription is the first step of almost every spoken-word edit, so we decided it should never cost you anything. In the Arsaze app for Windows, it now doesn't.
What changed
When you add a video in the Windows app, the speech in it is transcribed by your own processor. It is free on every plan, including the Free plan. It spends no AI credits, there is no per-minute charge and there is no hour allowance to run out of. We are not running a trial. This is how the desktop app works.
The recognition model is an NVIDIA Parakeet speech model that we run on your CPU, so you do not need a graphics card. The result is a transcript with a timestamp on every word.
Why we can make it free
Cloud transcription costs money every time it runs, which is why it draws on credits. When your own computer does the work, it costs us nothing to run, so we have nothing to charge for. We would rather give you that than find a way to meter it.
What you can do with a word-timed transcript
The timestamps are the useful part. A plain text file of what was said is a document. A transcript where every word knows its position in the video is an editing surface.
- Delete a sentence in the transcript and the video is cut at exactly those words.
- Strip filler words and shorten long pauses in one pass.
- Find the retakes of a line and keep the one you like.
- Add captions that follow the speech word by word.
- Search your footage by what people said.
These are the same tools that Rewrite uses. Because every one of them is also available to an AI assistant, you can ask Claude or ChatGPT to cut a ramble down to its best two minutes and it works from the same transcript you see.
What it does not do yet
We would rather you hear the limits from us.
- On-device transcription is English today. Speech in other languages is transcribed in the cloud, which uses AI credits where your plan includes them.
- It runs in the Windows app. In the browser, or on a Mac, transcription runs in the cloud, and the Free plan includes 50 one-time welcome AI credits for it.
- The app needs an internet connection while you edit, because your projects live in your account. Only the speech recognition step runs on your PC.
- We do not publish an accuracy percentage. It would depend on your microphone, your room and how people speak, and a single number would not describe your footage.
What happens to your audio
The speech recognition runs on your computer, so your audio is not sent to a transcription service. If you use a local project, your footage never leaves your PC at all. In a normal cloud project, the files you add are stored in your account as the project's media, as you would expect, but the transcript itself is made on your machine.
Try it
Download Arsaze for Windows, add a video and watch the transcript appear. The free transcription page has the short version, and the pricing page shows what every plan includes.
Questions people ask
Is the free transcription limited to a certain number of hours?
No. There is no trial and no hour allowance. Transcription on your own PC is free on every plan for as long as you use Arsaze.
Which languages can Arsaze transcribe on my PC?
English today. Other languages are transcribed in the cloud, which uses AI credits where your plan includes them.
Do I need a GPU?
No. The model runs on your CPU.
Does Arsaze transcribe in the browser too?
Yes, in the cloud. The Free plan includes 50 one-time welcome AI credits for it. The free forever version is the on-device one in the Windows app.