New Spektree's brother app — same roots

Cut the text. The video follows.

Speaktree transcribes your recording word by word, on your Mac. Strike a sentence and the cut is made. Then it finishes the job — captions, your brand, music, vertical clips, and the post that goes with them.

Get Speaktree Bundle with Spektree — save $90

macOS · Notarised by Apple · Transcription free and unlimited · Your video never leaves your Mac

live-ai-trainer.mp4 02:05:12

00:14 Okay. Here's the one idea I want you to leave with today: a model doesn't replace the trainer. it replaces the blank page. So you record the session once, and everything else follows from the words you said.

0:37 Click any word to cut it — click again to bring it back.
The recording nobody edits
“The live went brilliantly. The recording has been sitting in my Downloads for three weeks.”

A two-hour webinar holds maybe twelve minutes you would proudly post. They are buried between “can you see my screen?”, a silence while you hunt for a tab, and a question that went nowhere.

Finding them in a video editor means scrubbing a waveform for an afternoon. But you don't think in waveforms — you think in sentences. You remember what you said, not at which second.

Speaktree gives you the sentences. Every word timestamped, every pause visible, the strongest passages proposed. Editing becomes reading with a red pen — and the video quietly does what the text says.

From raw to posted

One recording in. Everything you publish, out.

01 · IMPORT

Drop it in, or film it here

Zoom, Teams, Loom, OBS, a camera file — or record your screen and yourself from inside the app, on separate tracks.

02 · READ

Edit like a document

The text appears while it transcribes. Strike what you don't want; start cutting before it has even finished.

03 · PROPOSE

Let it find the moments

The best passages, a montage by theme, the full demo — or just tell it what you want and it applies it.

04 · FINISH

Make it yours

Captions, your colours and logo, name and role, music, chapter titles — in one pass, at broadcast loudness.

05 · PUBLISH

Ship the whole kit

The video, vertical clips, a title, a description, YouTube chapters, a thumbnail and the post to go with it.

What it does

A video editor that reads first

Edit by text

Strike a word, the frame goes

Every word is pinned to its exact moment. Delete it in the text and the video cut is made — clean, with soft audio fades.

  • Cut words stay visible, struck through — nothing is destroyed
  • Search and cut every “um” in one gesture
  • Named versions: try an edit, come back to the last one
Silences

Pauses, priced before you click

Each silence preset tells you how much it would save — −3 min 08 s — before you commit to it.

  • Pauses shown in the text, where they happened
  • Keep the breath, lose the dead air
  • Undo is one click, always
Proposed by AI

It reads two hours so you don't have to

Claude reads the whole transcript and proposes what's worth keeping, in a column next to your text.

  • Best moments, a montage by theme, the full demo
  • “Three minutes, only the pricing part” — applied
  • A clickable outline of the whole session
Captions

Captions people actually read

Word-by-word karaoke captions, as big as you like, previewed live on the player while you set them.

  • Fix a misheard word once, for every project
  • Your own lexicon, shared across projects
  • Burned in, or exported as .srt
Vertical clips

A moment becomes a Reel

Any proposed moment turns into a 1080 × 1920 clip, captioned, ready for LinkedIn, Instagram, TikTok or Shorts.

  • Framing follows the speaker, with Apple's Vision
  • Screen-share on top, you underneath
  • A 40-second clip renders in about 14 seconds
Your brand

Finished, not just trimmed

Pick a brand kit and the whole video wears it — opening and closing cards, watermark, your name and role.

  • Loudness normalised for every platform
  • Punch-in zooms, transitions, colour looks
  • Chapter titles — or a mind map that draws itself
Music & sound

Music with no licence to check

Speaktree composes the soundtrack on your Mac, for your project. Safe on a client's video — there is no library to clear.

  • Ambiences from calm to driving, or bring your own
  • Ducks automatically under your voice
  • Whooshes and hits on the cuts, if you want them
Blur

Hide what the screen gave away

A client's name, a revenue figure, a private chat title. Draw a box on the player and it's gone from every export.

  • Mosaic, blur or solid black
  • Whole video, or just from one moment to another
  • Applied to the clips and the thumbnail too
Publish

The words that sell the video

Written for this edit — not the two-hour original — so the chapters land exactly where they should.

  • A title and two variants, a description
  • YouTube chapters timed to the edit
  • One post per video, and a 1280 × 720 thumbnail
Film it here

Your screen and you, on separate tracks

Record straight from Speaktree. Screen and camera land in two files, so you only appear in the video if you decide to — no green screen, no erasing yourself afterwards.

After the live

The recap your participants wanted

A summary of the session and the to-do list each participant left with, written from what was actually said. Send it while the room is still warm.

What comes out

The clip, and everything you need to post it

Speaktree doesn't stop at the export. One moment of your session becomes a captioned vertical clip in your brand — and the publishing kit is written from the edit itself, so the chapters and the post describe the video people will actually watch.

Publishing kit Written for the 3 min 34 edit, not the 2-hour live
Title AI won't replace trainers. It replaces the blank page. Variants · “The one idea from our AI-trainer live” · “Stop starting from zero”
YouTube chapters — timed to the edit
  1. 00:00Why the blank page wins
  2. 00:41Record once, teach many times
  3. 01:58What the model really writes
  4. 02:52Your first session this week
Post

Most trainers don't lack expertise. They lack a first draft. In this 3-minute cut from our live, the one shift that changes how you prepare a session…

Plus a description and a 1280 × 720 thumbnail · measured: $0.04, written in 23 s
Measured, not promised

A real two-hour webinar, on a Mac

Speaktree is new, so no testimonials yet. Here is a live webinar on the AI-augmented trainer — 2 hours 5 minutes, as it happened — put through the app, with the figures it wrote to the project folder.

Input A 2 h 05 live webinar, recorded straight through — slides, demos, questions and all
Transcribed
9m40
word by word, locally, $0
Dead air found
33 min
of pauses over 0.8 s
Moments proposed
$0.33
Claude, across all 2 hours
Vertical clip
14 s
to render 40 s, captioned

On a 15-minute course, applying one proposed edit took it to 3 min 30 — and the exported file landed within 40 ms of the length the editor announced.

Two trees, one root

Spektree makes the material. Speaktree makes the video.

Both start from the same thing: a recording of someone teaching on a screen. Spektree branches it into what you need to teach it — slides, trainer notes, chapters, a deck. Speaktree cuts it into what you need to publish it — the edit, the clips, the post.

Same principles underneath: it runs on your Mac, wears your brand, uses your own Claude key, and every project is a folder you own.

1 recording Slides Trainer notes Chapters Deck Spektree The edit Clips The post Speaktree
Question
You start with A video that taught you something — yours or someone else's Your own recording — or you film it in the app
You end with Slides, trainer notes, chapters, a branded deck, a guide A finished video, captions, vertical clips, the post
It's for the moment Before you teach After you've taught
Think of it as Your course, rebuilt Your video, finished

Prepare it with Spektree. Publish it with Speaktree.

Why local matters here

Your raw footage is nobody else's business

Raw recordings are where the mistakes live: the client dashboard you forgot to close, the private message that popped up, the half-finished thought. Uploading them to a cloud editor to cut them is exactly backwards.

Speaktree is an app, not a website. The video is read, transcribed, cut and rendered on your Mac. The only thing that ever leaves is transcript text, sent to Claude when you ask for proposals or a post — and you see what that costs, per project.

  • No upload of your video, ever. No account to create.
  • Transcription by local Whisper — free, offline, no hours to count.
  • Cutting, captions, clips, music and blur work with no key at all.
  • Exports are plain MP4 files in a folder in your Documents.
Pricing

Speaktree on its own, or the whole tree

No transcription hours, no export credits, no watermark on a paid licence. One price a year, and you keep everything you made.

Speaktree

$190 / year, per creator

Every feature, unlimited videos. Less than what an editor charges to cut one webinar — and you'll record more than one this year.

Get Speaktree Licence and download sent within 12 hours
Best value · both apps

Spektree + Speaktree

$390 $480 / year

Prepare with Spektree, publish with Speaktree. One licence for each app, one invoice. Already on Spektree? Add Speaktree for $100 / year.

Get the bundle Save $90 a year · founding members keep their price

Team bundle

$990 / year, 3 trainers

For training companies: both apps for three trainers — $1,180 bought separately. One named licence per person; each extra trainer $290 / year.

Equip your team Speaktree alone for a team: $490 / year for 3

Renew when the year is up, or don't: every video you exported is a plain MP4 and stays yours. In Morocco? Pay in dirhams — 1 900 MAD Speaktree, 3 900 MAD the bundle, by bank transfer via WhatsApp.

Every Speaktree licence includes

Signed and notarised by Apple: download, double-click, done. Your recordings never leave your Mac.

The AI parts run on your own Claude key — you pay Anthropic directly, at cost. Measured: about $0.04 for a publishing kit, $0.33 to read a two-hour webinar for its best moments.

  • Unlimited videos and unlimited local transcription
  • Text editing, silences, search-and-cut, named versions
  • Captions, vertical clips, blur, brand kits, music and sound
  • The publishing kit: titles, description, chapters, post, thumbnail
  • Filming from the app, and the after-the-live recap
Straight answers

Before you ask

Spektree or Speaktree — which one do I need?

Spektree if you are preparing to teach: it rebuilds the material of a session — slides, notes, chapters, a deck — from any recording. Speaktree if you have recorded and want to publish: it turns your own footage into a finished video, clips and a post.

Most trainers end up doing both, one after the other. That is the whole reason the bundle exists.

Does it replace Premiere or Final Cut?

No, and it isn't trying to. Speaktree is for videos where someone talks — webinars, courses, screen demos, interviews — and it edits them by what was said. If your video is B-roll cut to music with no words, a timeline editor is the right tool.

Do I need a Claude API key?

Not to edit. Transcription, cutting, silences, captions, clips, branding, music and blur all work without one. The AI features — proposed moments, the publishing kit, chapters, the live recap — use your own key, so you pay Anthropic what your videos actually cost, usually a few cents.

Is the music really safe on a client's video?

Yes. It isn't a library of tracks with licences to read: Speaktree composes it on your Mac, from scratch, for your project. Nothing to clear, nothing to credit. You can also bring your own track or your brand's music.

Which languages does it understand?

Whisper detects the language on its own — English, French, Spanish, Arabic and dozens more. Talks that switch languages mid-sentence come out less complete; any misheard word can be corrected in the captions, and the correction is remembered for your next videos.

How long does transcription take?

On a recent Mac, under a tenth of the video's length — the two-hour webinar above took under ten minutes. And you can start cutting as soon as the first sentences appear. (On battery in Low Power Mode it is much slower: plug in for long sessions.)

Is there a Windows version?

Not yet. Speaktree uses macOS for things like framing the speaker in vertical clips. If you need Windows, say so when you write — every request moves it up the queue.