Nvidia's Free AI Transcriber Is Out - How to Make Subtitles on Your PC Today

Nvidia has released a free AI model that turns recorded speech into text in 40 languages, but it is built for servers, not your laptop.

The good news: you can already make subtitles for your own videos with free AI, on your own PC, today.

It takes a free subtitle editor and a few clicks, and nothing gets uploaded anywhere.

Here is what Nvidia launched, who it is really for, and how to get subtitles for your videos right now.

What Nvidia Released

The new model is called Nemotron 3.5 Transcribe, and Nvidia released it on 28 October, TechCrunch reports. The model itself is free to download, including for commercial use.

• 40 languages. Including English, Spanish, French, German, Hindi, Arabic, Chinese, Japanese and most European languages - Irish is the notable gap.
• Long recordings. It handles files up to an hour long and can tell several speakers apart in some languages.
• Translation built in. It can turn speech in many of those languages into English text, and English into them.
• Recordings only. It works on finished audio files, not live streams or calls.

Nvidia says it beats rival models on public accuracy tests. Those are Nvidia's own figures, and independent tests will follow.

Can You Run It at Home?

Not really, for now. Nvidia's reference setup is a data-centre graphics card with 80 GB of memory, and the only home graphics card on its official list is the RTX 5090.

There is also no app with a button to press - it is a building block for developers. Expect it to appear inside apps and online services over the coming months rather than as something you install yourself.

How to Make Subtitles on Your PC Today

Subtitle Edit is a free, open-source subtitle editor that can listen to a video and write the subtitles for you. It uses speech engines that run on your own PC, including OpenAI's Whisper and Nvidia's smaller Parakeet models.

• Install Subtitle Edit and open your video. Drag the video file into the window, or open it from the Video menu.
• Choose Speech to text. Pick an engine - Whisper is a good start - and the language spoken in the video.
• Let it download the model. The first run fetches the engine once, then it works offline.
• Generate and save. Save the result as an .srt file with exactly the same name as the video, in the same folder.

Then open the video in VLC Media Player - it loads a matching .srt file automatically. If the lines run a little early or late, our guide to fixing out-of-sync subtitles sorts it in a minute.

Subtitles look different from player to player. If you want sharper styling or bigger text, see which player is best for subtitles - PotPlayer is a strong pick for styled ones.

Quick questions