Free Local Auto Captions for Premiere Pro (Whisper)
Add free auto captions to Premiere Pro with Whisper running on your own computer: 99 languages, no upload, no per-minute fees. Step-by-step setup for 2026.
Want automatic captions in Premiere without paying per minute or uploading your footage? This guide shows how to caption in Premiere with Whisper running locally, using the free Suflo panel.
Why a local Whisper panel?
Premiere Pro has its own Speech to Text, and for the languages it supports it works well. Two things push editors to look elsewhere: the language list is limited (see Adobe's supported languages), and many web caption tools want you to upload your footage and pay per minute. Suflo runs OpenAI's open Whisper model through whisper.cpp on your own computer, inside Premiere, for free.
| Suflo local engine | Suflo cloud engine | |
|---|---|---|
| Where audio goes | Stays on your computer | Sent to Groq for transcription |
| Needs | One-time engine download | A free Groq API key and internet |
| Speed (83 s of audio) | 8.5 s on CPU · 1.2 s with an NVIDIA GPU | A few seconds |
| Cost | Free | Free within Groq's free tier |
Speeds are Suflo's own measurement on one machine; yours will differ.
Free auto captions in Premiere: step by step
- Install Suflo from the latest release and open it from Window › Extensions (Legacy) › Suflo (Premiere 25.6+ and 2026) or Window › Extensions › Suflo.
- Choose English (beta) in the first-run guide and let it install the engine. It picks a small, fast model for your language; you can switch to Turbo or Large v3 later for fewer mistakes. Since Suflo 3.1 the guide also lets you try the engine on a sample clip and preview a caption style before you touch your own footage.
- Select what to caption: a clip on the timeline, the In–Out range or the whole sequence.
- Set the language or leave it on Auto, choose a line length (2–5 words or 32/42/60 characters) and press Generate captions.
- Fix and apply: correct names in the editor (add them to the glossary so they stay fixed), then apply to a caption track or export SRT, WebVTT or TXT.
Tips for better accuracy
- Use a larger model for difficult audio, accents or music beds.
- Add brand names and jargon to the glossary as
wrong => rightrules. - Leave silence skipping (VAD) on: it ignores parts without speech and reduces made-up captions.
- Clean noisy audio first; Suflo Pro's Enhance audio keeps sync while it denoises.
What Suflo does not do (yet)
- It is a Premiere panel only: no DaVinci Resolve or Final Cut.
- Animated, burned-in caption styles are part of Suflo Pro; the free tier gives you plain caption tracks and files.
- Styled caption fonts cover Latin, Cyrillic, Greek and Arabic scripts; for Chinese, Japanese, Korean and Indic scripts use the caption track or SRT.
Sources
- Whisper model and language list: github.com/openai/whisper
- whisper.cpp: github.com/ggml-org/whisper.cpp
- Premiere Speech to Text languages: Adobe Help
Adobe and Premiere are trademarks of Adobe Inc. OpenAI and Whisper are trademarks of OpenAI. Suflo is not affiliated with or endorsed by either company.
Local auto captions for Premiere in 99 languages, plus one-time-price Pro editing tools. No subscription, no credits, no account.
Download free → See ProFrequently asked questions
Is Suflo's captioning really free?
Yes. Transcription, the caption editor, SRT/VTT/TXT export and applying captions to a Premiere caption track are free and open source (MIT), with no time or minute limit.
Does it work offline?
Yes. After the one-time engine download, the local engine needs no internet, account or key. The optional cloud engine (Groq) needs internet.
Which model should I use?
Start with the model the first-run guide suggests. Larger models such as Turbo or Large v3 make fewer mistakes but are slower; an NVIDIA GPU makes every model much faster.
Can I edit the captions before they go on the timeline?
Yes. Suflo opens every transcript in its own editor: split, merge, retime, shift and fix words, then apply to a caption track or export SRT/VTT.