Blog

The free b-roll and stock footage plugin for Premiere Pro

11 October 2026 · 6 min read

Summarize withChatGPTClaudePerplexity

If you make long voiceover videos, such as documentaries, explainers, history or space channels, or sleep stories, you know the part that takes longest isn't the voiceover. It's the footage. Every sentence needs something on screen, so you open a stock site, search, preview, download, drag the clip into Premiere, trim it to the sentence, and do it again. For a two-hour video that loop runs more than a thousand times.

We built VoxVision to take that loop off your hands: a free b-roll and stock footage plugin that works inside the Premiere Pro timeline, made for videos that run for hours.

The real problem with b-roll in long videos

Finding one good clip is easy. Finding twelve hundred, in order, cut to the right length, without the same galaxy shot every few minutes, is not. Long videos also show small mistakes: a clip that ends one frame late is invisible once, but after a thousand clips the footage is seconds behind the voice. And at the end, someone has to write down where every clip came from.

Why most stock footage tools stop halfway

A stock site is a search box. It finds the clip, and the rest is still yours: downloading, conforming it to your sequence, trimming it, placing it and crediting it. Tools that search from inside Premiere save the tab switching, but you still place clips one at a time. That is fine for a ten-minute video. For a two-hour one, the whole chain has to happen by itself, from the voiceover to the finished timeline.

How VoxVision works

It reads the voiceover

VoxVision transcribes the voiceover with Premiere Pro's own Speech to Text, or with a speech model it installs on your PC. Your audio never leaves your computer. It splits the words into sentences, and each sentence becomes a spot on the timeline.

It finds what each sentence can show

A sentence like “Our Sun is one of two hundred billion stars” becomes search words a camera can film: sun, stars, milky way. Abstract sentences fall back to footage that fits the topic of the whole video, so no spot is left empty.

It searches and filters before anything downloads

It searches Pixabay, and Pexels if you have a key, with your own free API keys. Clips with text, logos or watermarks, the wrong orientation, too low a resolution or too short a length are left out first, and near-duplicates are removed by comparing preview images. Only then does it download, and only the clips it will use, each in the smallest size that still covers your sequence.

It cuts and places the clips

Every clip is cut to the exact number of frames of its sentence and placed on VoxVision's own tracks, with the stock audio muted. Your voiceover, your facecam and any clip you placed yourself are never moved, cut or muted. You can replace any clip with another match, block footage you never want to see again, and copy a credits list for the video description.

Built for videos that run for hours

• Batches: it searches, downloads and places a few clips at a time, so the first shots are on your timeline while the rest of the video is still being found.
• Stop and continue: press Stop, look, and press Continue. It carries on from the first empty spot.
• No clip over and over: each clip is used at most three times, at least ten minutes apart, from a different moment of the clip.
• In sync to the last frame: shots are counted in whole frames from the start of the voiceover, so nothing drifts in hour three.
• Three paces: a new clip every sentence, whole clips that play in full, or short b-roll over your own video.

More on this in how to add b-roll to a 2–3 hour YouTube video and on the ultra long-form page.

What you need

• Windows 10 or 11 and Adobe Premiere Pro 2026, version 26.3 or later.
• A free VoxVision account and an internet connection.
• A free Pixabay API key. Our step-by-step guide shows where to find it.
• No graphics card is needed.