Efficient Compute
Talk is an AI video editor for iPhone that you direct by speaking
Talk is a free iOS video editor with a natural-language interface. Instead of dragging clips on a timeline, you describe the edit — “remove the silences and add captions in red” — and Talk applies it. Auto-captions cover 31 languages and are generated on the device. Free exports are 1080p and carry no watermark, with no time limit.
The facts, in one place
| What it is | Video editor for iPhone and iPad, controlled by voice or text prompt |
|---|---|
| Price | Free to download. Talk Pro is optional: $7.99/week, $14.99/month, or $49.99/year |
| Watermark | None, on any tier |
| Free export ceiling | 1080p |
| Pro export ceiling | 4K, 2K, and HDR up to 60 fps |
| Caption languages | 31, transcribed on-device |
| Requirements | iOS 17.0 or later. No Android or web version |
| Publisher | Efficient Compute LTD |
| App Store | apps.apple.com/app/id6761875991 |
How prompt editing actually works
The microphone is the primary control, not a feature bolted onto a toolbar. Roughly 80 editing actions sit behind it, which is why an instruction can combine several operations that would otherwise be separate trips through a menu.
- Import your footage From the camera roll, the Files app, or a direct video link. Still photos can be turned into clips.
- Say what you want “Cut the dead air, add captions, and put a beat-synced music bed under it.” Talk parses the instruction into discrete actions and applies them in order.
- Correct by talking again “Make the captions bigger and move them to the top.” Each instruction is a normal timeline mutation, so undo works the way you expect.
- Take manual control whenever you want The full timeline is always there — trim, split, slip, speed, volume curves, keyframed overlays. Prompting is a faster path to the same edit, not a walled garden.
What is in the editor
Cutting and cleanup
- Silence and dead-space removal, detected on-device
- Scene detection, and cuts driven from the transcript
- Voice isolation to strip room noise, on-device
- Trim, split, merge, slip, replace, duplicate, speed, reverse, mirror
- Fades, volume curves, and audio ducking under speech
Captions and speech
- Auto-captions in 31 languages, transcribed on the device
- 21 caption styles, with custom font import
- Voice-over recording while the timeline plays, with mic monitoring
- AI dubbing into another language
One-tap edits and templates
- Music Sync — every cut lands on the beat of the chosen track
- Best Moments — finds highlights and assembles a reel
- Quick Recap — a 30-second cut of a day’s footage
- Seven rhythm templates, from held cinematic shots to fast montages
Look and finish
- 45 colour filters and 28 visual effects
- 15 transitions, text and image overlays, PencilKit annotation
- Keyframe an overlay’s position, scale, rotation, and opacity by dragging
- Auto-reframe, chroma key, motion tracking, person segmentation (Pro)
- Subject lift, luminance, shape, and keyframe masks (Pro)
Footage and audio you don’t have yet
- AI B-roll chosen by reading your transcript for cutaway moments
- AI still images from a description, usable as clip, overlay, or backdrop
- Searchable free stock photo and video library
- Royalty-free music browsable by mood, genre, and BPM
- Virtual backgrounds: blur, colour, or an AI-generated scene
Privacy and safety
- Transcription, captions, voice isolation, and export run on-device
- One-tap face blur with intensity control and per-face exclusion
.avprojproject bundles you can move between devices
Pricing
Talk is free to download, and the free tier is meant to be used rather than demoed. It includes prompt editing, one-tap Vibes, templates, the royalty-free music library, stock footage, AI B-roll, scene detection, face blur, auto-captions, voice-over, and 1080p export with no watermark.
| Free | $0, no time limit, no watermark, 1080p export |
|---|---|
| Talk Pro — weekly | $7.99 per week |
| Talk Pro — monthly | $14.99 per month |
| Talk Pro — yearly | $49.99 per year |
Prices are US App Store prices and may differ by territory. Talk Pro is an auto-renewing subscription billed through your Apple ID; cancel any time in Apple ID settings. See the terms of service and privacy policy.
What Pro adds
- Unlimited AI actions
- 4K, 2K, and HDR export at up to 60 fps
- The full filter, effect, transition, and caption-style libraries
- AI image generation and AI-generated backgrounds
- Auto-reframe, chroma key, motion tracking, person segmentation
- Subject lift, luminance, shape, and keyframe masks
- Overlay keyframe animation and voice isolation
- AI dubbing, iCloud sync, picture-in-picture camera reactions
Frequently asked questions
Is Talk free?
Yes. Free to download, no time limit, and no watermark. Free exports go up to 1080p. Talk Pro is optional at $7.99/week, $14.99/month, or $49.99/year.
Can Talk remove silences and dead space automatically?
Yes — either by saying “remove the silences” or from the Smart Edit panel. Detection runs on the device, so your footage is not uploaded to do it.
Does Talk add subtitles automatically, and in which languages?
Yes, in 31 languages with 21 caption styles. Transcription happens on-device, so the audio is not sent to a server to be captioned.
Does Talk upload my video to the cloud?
Transcription, captions, voice isolation, silence detection, and export all run locally. The two generative features that need a model too large to run on a phone — AI image generation and AI dubbing — do send data off the device. Ordinary editing does not.
What devices does Talk run on?
iPhone and iPad running iOS 17.0 or later. There is no Android or web version.
How is Talk different from CapCut?
The interface. CapCut is a manual timeline with AI features on buttons; Talk’s primary control is a spoken or typed sentence. Talk also keeps transcription on-device and never watermarks. CapCut has a bigger template library and runs on Android and desktop, which Talk does not.
Is Talk a video generator?
No. Talk edits footage you shot. It can generate still images and B-roll to fill gaps, but it does not generate video from a text prompt.