The public record of every Caption Plug release - what's new, what got better, what got fixed. Updates install themselves from inside the plugin (the ⬆ Update button), and every v1.x release is included with your one-time purchase.
Current build: v1.5.7
v1.5.7
Box colours and live preset previews
Choose the colour of any box, pill, or bar, see every preset animate in your own style before you pick it, and caption with Premiere's built-in transcriber.
NewPresets that paint a box, pill, sticker, censor bar, marker, card, or badge now have their own colour control in Customisation. Each preset remembers its colour, and nothing changes until you pick one.
NewMulticolour Box: the jumping box changes colour with every caption, cycling through four colours you choose.
NewEvery preset card shows a small looping preview in your current font, colours, stroke, and shadow, so you see exactly what it will put on your timeline.
NewPremiere Pro Speech to Text is available as a transcription option. Create captions in Premiere, export them as SRT, and Caption Plug turns them into animated captions. Premiere's word timings are less precise than Whisper's.
ImprovedHormozi Box now uses one colour of your choice instead of changing colour caption by caption. The cycling look lives on as Multicolour Box.
ImprovedRendering is faster and uses less disk space: frames that don't change are no longer encoded again, and the spare frames after back-to-back captions are shorter.
FixedThe preview keeps the true shape of your sequence. Enlarging it could stretch the picture, especially for vertical videos.
FixedReinsert Captions puts back only the captions that are actually missing, even if you moved the rest to another track, so it can no longer create a duplicate layer.
v1.5.6
Caption Sets: transcripts that outlive your timeline
Every generation becomes a durable Caption Set you can reopen and edit at any time, captions can no longer overwrite your footage, and local transcription works on AMD and Intel Windows machines.
NewEvery generation becomes a Caption Set - a durable record of its transcript, timings, style, and rendered captions - that stays editable no matter how you move, trim, re-cut, or delete the caption clips on your timeline.
NewEdit Captions, formerly Edit Transcript, follows your timeline selection. Select captions from several different generation runs and they appear as one transcript in playback order, each block labelled with the run it came from.
NewWord-level timings are editable underneath each caption block - click the timestamp to open them. Correcting a word keeps every timing intact, and nothing is ever retranscribed.
NewReinsert Captions puts rendered captions back on the timeline after you delete them, without retranscribing or re-rendering.
NewCaption Sets are saved with your Premiere project, so they survive a restart, a panel reload, and a Caption Plug update. Transcripts from earlier versions are imported automatically on first launch.
ImprovedApply Changes now sits at the top of the editor at all times, greyed out until there is something to apply, and re-renders only the captions you actually edited across every Caption Set on screen.
ImprovedThe editor no longer clears itself. Deselecting clips, or selecting something unrelated in the timeline, leaves your work exactly where it was.
FixedCaptions can no longer overwrite your footage. Caption Plug places them on a genuinely empty track, adds a track when every existing one is in use, skips locked tracks, and refuses to place captions that would land on your own media.
FixedLocal transcription on Windows now runs on AMD, Intel, and CPU-only machines, which could still fail to start after the previous release. NVIDIA machines keep GPU acceleration and fall back automatically if it fails.
FixedDropping a file onto the panel no longer replaces it with a file-not-found error that could only be cleared by reopening the extension.
FixedArabic, Hebrew, Persian, Urdu, and other right-to-left captions render in reading order, keep their cursive joins on per-character animation presets, and run directional effects such as karaoke fills and marker swipes in the reading direction.
v1.5.4
Local transcription without an NVIDIA GPU
Windows machines without a working NVIDIA GPU can transcribe locally instead of failing to start.
FixedLocal Whisper on Windows starts in CPU mode when no working NVIDIA GPU is detected, so AMD, Intel, and CPU-only systems can transcribe locally.
ImprovedNVIDIA systems keep CUDA acceleration, with one automatic CPU retry if GPU startup fails or times out.
ImprovedSettings only reports NVIDIA acceleration after a successful driver check, and drops that claim after a fallback to CPU.
v1.5.3
A font picker that opens every time
The font dropdown in Customisation no longer intermittently refuses to open.
FixedThe font list now renders inside the panel instead of relying on a native popup window, so it opens reliably every time.
ImprovedGrouped bundled fonts, imported fonts, custom font names, and keyboard selection all behave as before, and the list fits above or below the control within the panel.
v1.5.2
Custom highlight text colors and a reliable Windows updater
Choose the text color used on highlight boxes and marker strokes, with the pending Windows updater reliability fix now included for everyone.
NewA dedicated Text on highlight color control now styles words inside Hormozi Box, Box Trail, Marker Swipe, and related solid highlight presets.
ImprovedNamed visual styles save and restore the highlight text color independently from the regular text and highlight colors.
FixedBox Trail now keeps the selected text color on every retained highlight box instead of reverting earlier words.
FixedWindows updates wait for the download to finish closing before extraction and use Windows' built-in archive support for the larger accelerated runtime.
v1.5.0
Multi-Speak, hands-on styling, and safer workflows
Detect two-person speaker turns locally, shape captions directly in the preview, reuse your own styles and fonts, and cancel long jobs without leaving a partial timeline edit.
NewMulti-Speak (Beta) detects turns in two-person dialogue after any transcription provider, splits captions at each speaker change, and gives each speaker a separate accent color.
NewLocal speaker analysis uses an optional, checksum-verified 102 MB download. Audio stays on your computer, the existing WAV export is reused, and the components can be removed any time.
NewSpeaker names and accent colors are editable. Names stay inside Caption Plug; Animated output varies the selected preset's accent, while Standard Captions keep speaker-turn grouping with Premiere-managed styling.
NewIf speaker detection is uncertain or fails, Retry detection and Continue without speakers reuse the completed transcript instead of transcribing the audio again.
NewSave up to 40 named visual styles and restore their animation, typography, colors, shadow, spacing, placement, and words-per-caption settings without changing provider, API key, SFW, or account settings.
NewImport OTF, OTT, and TTF fonts for Animated output. Caption Plug keeps a private copy, adds it to the Imported font group, and reloads it after panel restarts.
NewMove and resize captions directly in the live preview by dragging, using the corner handle, or pressing the arrow and plus/minus keys.
NewHorizontal placement, a wider 1% to 24% size range, and full 0% to 100% position controls make caption layout more flexible while safe-frame clamping keeps the result visible.
NewEnlarge the live preview inside the panel or open it full screen. The enlarged preference is remembered.
NewShadow distance and softness controls now work alongside color and angle across word captions, stickers, pixel effects, chrome, title cards, and preset-specific rendering.
NewA Keep punctuation option restores sentence punctuation when wanted. New transcripts omit it by default, and each choice keeps a separate cache.
NewClick a transcript timestamp to edit that word's exact start and end time, then apply the timing correction with the same selective Apply Fixes workflow used for text edits.
NewCancel current task now stops transcription, uploads, local processes, speaker detection, keyword selection, rendering, SRT work, censoring, and caption replacement safely.
NewAnimated caption media now includes two seconds of held tail frames, so a caption clip can be extended in Premiere without exposing a blank or restarting the animation.
ImprovedStyle controls are reorganized into Saved styles, Typography, Layout, Colour and outline, and Shadow sections.
ImprovedLong captions can wrap beyond the previous two-line limit, oversized words shrink to fit, and every layout remains clamped inside the frame at larger sizes and off-center positions.
ImprovedFast / Whisper Base now requests word-sized whisper.cpp segments while retaining DTW alignment, producing tighter word timing.
ImprovedManaged Local Whisper on Windows now includes CUDA acceleration for compatible NVIDIA GPUs, with the required runtime files bundled in the installer.
ImprovedCloud transcription uploads stream from disk instead of loading the entire audio file into panel memory, and cancelling cleans up the upload and temporary file immediately.
ImprovedSFW mode now follows the transcript language for English, Spanish, French, and German. Auto-detect checks all four dictionaries and accented words are handled correctly.
ImprovedCaption timing holds the final word long enough to reach a rendered frame, fills short gaps, and never overlaps the next caption.
ImprovedCaption Plug reuses the nearest unlocked, non-overlapping video track above visible media and creates a new top track only when no existing track is safe.
ImprovedTranscript memory now retains 64 recent records, keeps full render details for the newest 8, and recovers the last complete disk backup after an interrupted save.
ImprovedCaption attachments now follow razor-cut source pieces independently, reconnect after an undo-restored clip, survive nested or linked A/V replacement, and ignore unrelated pasted copies.
ImprovedSpeaker identity now stays attached through transcript edits, dead-air removal, SFW censoring, keyword titles, multi-clip selection, saved transcript memory, and rendering.
FixedSelective Apply Fixes preflights the full replacement and restores the original batch if any change fails, preventing a half-updated timeline.
FixedWord timing edits preserve the existing caption grouping, recalculate neighboring holds, and insert only the intended duration so the next caption is never overwritten.
FixedCancelling animated generation removes every caption clip from the partial run while leaving unrelated timeline clips untouched.
FixedVery short or crowded SRT cues now keep proportional word timings monotonic and non-overlapping.
FixedIn-panel updates validate a complete release before swapping folders and automatically restore the previous version if installation fails, preventing mixed-version panels.
FixedUpdates stop the managed local runtime first, clean partial downloads and staging data, and refuse to replace a developer checkout.
v1.4.2
Animated SRT imports and precise multi-clip editing
Bring existing subtitles straight into the animation workflow, control SFW exceptions, and keep batch-generated captions attached to the right source clip.
NewImport an SRT subtitle file and turn its timed cues directly into animated Caption Plug captions.
NewAdd exact words to a censoring white list so SFW mode leaves those words visible and unbleeped.
ImprovedMulti-clip caption runs now remember each source clip separately, so selecting, editing, or moving one clip affects only its own transcript and captions.
FixedCaption grouping no longer joins words across separate source clips in the same batch.
ImprovedThe Fast / Whisper Base model now includes a clearer accuracy note, and the video tutorial can be opened directly from Settings.
v1.4.0
Windows updater fixed, shadow controls, and boundary protection
Windows users can now update from the panel, captions stay inside the frame, and you control shadow direction and color. Special thanks to Damir and Sebastian.
FixedWindows updater now works—PowerShell script argument passing has been fixed so the Update button successfully installs new versions.
FixedLarge captions now stay within the frame automatically, with boundary checking that prevents overflow beyond the top or bottom edge.
NewBasic Highlight preset added to Boxes & Highlights—the whole sentence appears instantly with only the active word changing color, no entrance or exit animations.
NewDrop shadow color picker lets you choose any shadow color instead of only black.
NewShadow angle control (0-360°, 15° steps) sets the shadow direction for all caption styles, applied to word captions, stickers, pixel art, chrome effects, and title cards.
v1.3.9
Frame-accurate captions that stay attached
Local captions land on exact sequence frames and follow only the original audio or video clip. Special thanks to Damir for the Windows updater find and Sebastian for the Preset Suggestion.
FixedManaged Local Whisper now keeps its word-alignment data, producing consistent word timing across the full transcript.
FixedAnimated captions now sample and land on exact Premiere sequence frames instead of drifting early or late.
ImprovedCaptions follow their source automatically after a timeline move, without selecting the clip again.
FixedEach caption run stays attached to its original timeline clip, so moving a pasted copy no longer pulls the original captions with it.
FixedMoving a source clip backward now relocates its captions cleanly without leaving duplicate clips behind.
ImprovedSource tracking checks attached caption runs together and only performs timeline work when a clip actually moves.
v1.3.8
Captions that stay with the edit
Animated captions now follow source moves, accept precise timing correction, and make transcript cleanup faster.
NewAnimated captions now follow their selected source clip when it moves in the Premiere timeline.
NewA saved timing-correction control can move generated caption timing up to one second earlier or later.
ImprovedClearing a word in the transcript editor now removes it from the affected animated caption instead of blocking the selective fix.
NewSigned-in customers can leave an account-verified rating from the panel and open the bug or idea form from Settings.
FixedLarge V3 local transcription now uses the matching word-alignment preset, while the smaller model is labeled accurately as Fast / Whisper Base.
v1.3.7
Transcript memory and a cleaner panel
Caption Plug now remembers generated transcripts when the panel is reopened, follows selected timeline audio, and offers a more focused three-workspace layout.
NewManaged Local Whisper now offers Turbo and Large V3 as independently downloadable models, with clear install, verification, selection, and removal controls.
ImprovedThe panel is organized into Customisation, Edit Transcript, and Settings workspaces while keeping caption generation and progress visible.
ImprovedAfter captions are generated, selecting their audio or video clip opens the matching timestamped transcript automatically.
FixedGenerated transcript records, pending word edits, and selective caption-fix details now survive closing and reopening the plugin panel.
FixedAudio that has never had captions now offers Generate captions; Regenerate captions is reserved for captions that were previously created and later removed.
ImprovedThe remove-all filler-word action is now a smaller control beside Select all or Clear all in the reviewed speech-cut workflow.
FixedCaption Plug now rejects stale selection ranges and sequence changes before output can be placed at the wrong timeline location.
v1.2.0
Transcript editing and reviewed speech cuts
Edit timestamped words, remove reviewed pauses and hesitation sounds, and fine-tune caption spacing without leaving Caption Plug.
NewA Transcript & cuts workspace shows every transcribed word with its timeline timestamp and supports word-level corrections.
NewAfter animated captions are generated, transcript corrections can replace only the altered caption clips on the timeline.
NewCaption Plug can propose significant pauses, high-confidence hesitation sounds and optional common filler phrases for review, then ripple-cut the selected ranges and stitch the selected dialogue clips together.
NewLetter spacing and word spacing controls make cramped captions easier to read, and the customization panel now has a one-click reset to defaults.
ImprovedThe live render preview stays pinned while you scroll and now reflects stroke width, letter spacing and word spacing changes immediately.
FixedSigning in again on the same computer now reuses one persistent device identity instead of consuming another machine activation when network hardware changes.
FixedThe website navigation now identifies the signed-in account instead of continuing to show Log in.
v1.1.3
Trimmed audio keeps the right caption range
Captions now follow the audible section of a moved, trimmed clip instead of reusing its old source range.
FixedMoving a trimmed audio clip now recalculates the caption range from its current timeline position and source in-point, preventing stale-range captions and [Blank Audio] results.
v1.1.2
Captions now follow the edit
Selected clips are automatic, moved clips keep their timing, and Local Whisper gets tighter word alignment.
NewLocal Whisper is now fully managed by Caption Plug. Download the 141 MB multilingual model once, then transcribe on-device without an API key, external server or internet connection.
FixedGenerate, SRT export and audio censoring now read the live timeline selection every time. Move the same clip and its cached transcript follows to the new position automatically.
FixedSelecting a video clip without its linked audio no longer falls back to zero seconds. Caption Plug uses the audible dialogue in the selected picture range.
ImprovedLocal word timing now uses a stronger speech-alignment pass, and animated highlights sample the exact sequence-frame time for tighter sync.
FixedLong caption runs stay together on one video track instead of creating another track for each batch.
ImprovedA run is pinned to the sequence where it started, and its settings lock until completion, preventing output from landing in the wrong sequence or changing style halfway through.
v1.1.1
In-panel updates repaired
This maintenance release fixes the installation-path issue that prevented Caption Plug 1.0 from applying its first update.
FixedCaption Plug now converts Premiere's encoded extension URL into the correct macOS or Windows filesystem path before installing an update.
ImprovedUpdate installation now remains compatible with the older CEP runtime included with Premiere Pro 2020.
ImprovedIf you are currently on version 1.0, install 1.1.1 once from your Caption Plug account page. Future updates can then be installed directly inside the plugin.
v1.1.0
Whisper can now run entirely on your device
Caption Plug now connects directly to local Whisper, so captions can be transcribed without a cloud API key or internet connection.
NewLocal Whisper is now a transcription provider. It works with whisper.cpp on its standard port and with OpenAI-compatible local transcription servers, while keeping word-level timing for animated captions.
ImprovedCaption Plug prepares Premiere audio for whisper.cpp automatically when ffmpeg is available, and gives clear setup guidance when the local server cannot be reached.
FixedSwitching transcription providers, languages, endpoints or local models now creates a fresh transcript instead of reusing a cached result from the previous setup.
FixedThe panel header now stays clear and readable at Premiere's narrowest supported panel width.
v1.0.0
One-click installers and account sign-in
The install and activation experience got a full overhaul the day after launch - installing Caption Plug now feels like installing any other app.
NewNative installers for both platforms - CaptionPlug-Setup.exe on Windows and CaptionPlug.pkg on macOS. Download from your account page, double-click, next-next-done: the right Premiere Pro folder is preselected and no admin rights are needed.
NewActivation is now just signing in. Open the panel and enter your captionplug.com email and password - no license keys to copy. One purchase still covers 3 machines, managed from your account page.
NewCensor audio only - bleep profanity on your timeline without generating captions at all. One button transcribes, finds the words, and drops the bleeps and dialogue cuts in place.
NewThe 62 bundled caption fonts are installed for Essential Graphics automatically, so Editable text mode can use them too.
ImprovedGenerate now takes audio straight off your timeline by default - select just your dialogue clips for the cleanest transcript (music beds are muted during the export), or select nothing to caption the whole sequence.
FixedAccount downloads always serve a working build: if a platform installer isn't available for a release yet, the download falls back to the universal zip instead of failing.
v1.0.0
Caption Plug 1.0 - initial release
The first public build: animated, frame-accurate captions rendered directly onto your Premiere Pro timeline. One purchase, yours for life.
New45 animated caption presets across seven families - Beast Pop, Hormozi Box, TikTok Pill, karaoke fills, glitch, neon, typewriter, earthquake and more. Every preset is a mechanically different animation, not a recolor.
New8 keyword-title presets for long-form: the panel picks the key moments out of the transcript - names, numbers, punchy statements - and drops a clean animated title on each.
New62 bundled caption fonts, grouped to match the preset families - zero font installation needed for animated captions.
NewTwo output modes: Animated clips (transparent PNG-sequence clips at your sequence's exact resolution and frame rate - frame-accurate by construction) and Editable text (a native Premiere caption track that stays editable).
NewSFW mode for brand-safe edits: censors profanity in captions (F**k style), drops 1 kHz bleeps on a new audio track, and can gate the original dialogue under each bleep so only the bleep is audible.
NewWhisper transcription with word-level timestamps through your own OpenAI or Groq API key - most major languages, auto-detected or set explicitly.
NewPlain SRT export for YouTube and other editors, live preview running the actual render engine, a built-in color picker, and a guided first-run tutorial.
NewUpdates are built in: when a new build ships, an Update button appears in the panel's top bar - one click installs it in place, settings kept.
All of it, $14.99 once.
Every release on this page - and every v1.x release after it - is included with one purchase. No subscription, no upgrade tiers.