Starting a Project

When you click New Project, ScriptCut asks one question first: how are you working? Your answer tailors the whole experience, the import steps, the export options, and the tutorial tips, to your editing software.

The workflows

Blank Project

Start from scratch and import any supported transcript later. The most flexible option.

Transcribe

Upload audio or video and ScriptCut generates the transcript for you, word by word, with speaker detection.

DaVinci Resolve

Optimized for Resolve’s subtitle export and XML roundtrip. The import tips name the exact Resolve menus.

Adobe Premiere Pro

Optimized for Premiere’s transcript JSON and XML roundtrip.

Final Cut Pro & Avid Media Composer

ScriptCut works with both. What is coming soon is only their own start-here tile in this picker, with its guided roundtrip. Today you start from Transcribe or Blank and export the Timeline (XML), which opens in Avid and Final Cut like any other editor.

Social Clips (Social, ProAI and Team)

Skip the transcript workflow and go straight into the Social Clips suite: upload a video and start building short, vertical, AI-reframed clips. See the Social Clips chapter.

Several recordings in one project

A project can hold several recordings. The Import tab is where the files live: its Video / Audio step lists every video or audio file in the project, in order, each with its full file name, in a list under the drop zone, which stays on screen and reads Upload more files. You can drop several files at once, and nothing moves on until you press Next. The Import from an app button inside the drop zone lists your switched-on apps (Frame.io, Google Drive, Dropbox, Box, OneDrive, Zoom, Vimeo): a connected one opens its picker and a file you pick joins the list like a dropped one, an app that is not connected says Connect and opens Settings. Each file has its own progress bar, and Next stays grey until every file is Ready for transcription (its audio is out). The Recordings item in the ⋯ page menu jumps there.

After you add a file, Get Your Transcripts lists every file with a checkbox: ticked means ScriptCut transcribes it (minutes are charged per file and each file rounds up to the next full minute; the minutes you have are shown under the title), and Select all at the top ticks every one. Any file can instead get its own transcript with Add transcript on its row. You add them one file at a time, and ScriptCut does not match transcripts to files for you. Each file is one line, with its name, length and frame rate. Files that already have a transcript say Transcript ready. A file still uploading is transcribed as soon as its upload finishes, and its lines are added to the project when it is done. Every screen shows a header above each file’s lines with its full name and which file it is (for example File 2 of 3), and Cleanup always lists each file’s lines together, in file order. Hover a line’s timecode to see which file it comes from. In the list, use the arrows to change the order, or Remove to remove a file (its lines stay but can no longer be played); click a file’s name above its lines to open the Recordings window and rename it. A file with no transcript shows Needs a transcript in red with a Transcribe button. You can also bring your own transcript for each file: add it on that file's row in Get Your Transcripts, or with the upload button beside Needs a transcript in the Recordings window, and ScriptCut checks it against the file's length and timecode. Anything it is not sure about waits on a matching board, where you can play a few seconds of each file to decide. When several videos are transcribed together, ScriptCut waits until all of them are done: one Progress screen shows one overall percent and a row for each video, and a video that failed offers Try again (or Buy minutes) and Continue without it. Then every file’s lines land together, in file order, and a Name Your Speakers page lists the speakers of each video with one play button each. Type a name to name a speaker, or open the arrow beside the box and pick someone already named to merge them; a voice that sounds like an earlier one is pre-picked (“Sounds like Speaker 1”), but nothing merges until you press Continue. Who Is Who then lists each person once, with the roles. A project with one video skips Name Your Speakers. The same page opens when you add a file later, bring your own transcripts or merge projects, and Match speakers in the file’s header row or in Cleanup > Tools opens it again. You get one “ready” email for the whole project.

The timeline XML, the Edited Video download and the Edited Audio, High Res download all cut across every file, each file at its own frame rate and timecode. A timeline XML for each file works too: the XML step lists every file with its own Add XML, or drop several at once and each goes to the file whose media it cuts. Each file's lines are cut from its own timeline and the exported XML joins them into one, each file at its own frame rate (the timeline runs at the highest frame rate among the files, and at the first file's rate when two files tie). The Edited Video needs a picture in every file it uses (a sound-only file stops it, and says which), and burned-in subtitles are not available for an Edited Video made from several files. Social clips can mix lines from every file as well: in Chrome and Edge the preview plays straight through a change of file (it gets the next file ready a few seconds early); in other browsers, or for a line under about a second long, it can pause for a moment there (the downloaded clip never does), and a line from a file whose picture is a different shape is left out of that clip, with a note saying so. Only the quick Edited Audio MP3 cannot combine files: if a selected line comes from another file, it makes no file and says so. The timeline XML and the Edited Audio, High Res handle several files.

Recording Tags. When one host asks the questions in two places, their lines merge into one speaker, so in Arrange you could not tell which question belongs to which answer. Open the page menu (three dots) on Cleanup, Select or Arrange and set Recording Tags to Hide or back to Show (it starts on Show; the choice is remembered in your browser). Every file then gets a letter, A for the first file and B, C in the order files were added, and a color. The letter sits on the file's header, beside the speaker names on Cleanup and Select, and on every line in Arrange; hover it to see the file's name. A letter never changes when files are moved or another is removed. Click the letter on a file's header to give the file a short name (for example Studio or Rooftop) and pick its color. In Arrange, when a question from one file sits right before an answer from another, a short amber note between them says Question from A, answer from B; its X hides it for that pair. Two interviewees from two files back to back never warn. Keepers stay green and maybes yellow; the tags never change them. The Review page never shows tags.

The AI tools and several files. AI Select, AI Story, AI Arrange and AI Clips are told which file every line comes from. When the same answer is in two files (two takes, or two cameras on one moment), the AI keeps only the strongest telling, never both, and a short stretch or a Trim ends only clip stays inside one file. While a file's speakers are not matched yet, every AI window shows a notice with a Match Speakers button: an interviewer tagged in one file is a stranger in the next until the two are matched. The AI never picks a Crew line in any module, so tag slate and crew talk as Crew in Roles.

Several uploads in AI Story: Keep separate or Mix. When a project holds more than one upload, the AI Story window shows a Several uploads choice. Keep separate builds each upload as its own part of the story and never cuts a line from one upload next to a line from another. Mix lets moments from different uploads meet on a theme. ScriptCut picks the default from who is in each upload (different people = Keep separate, the same people in every upload = Mix) and you can flip it before running; the choice is remembered with the project. In both modes a reply never follows a line from a different upload: the real answer from the same upload is placed after a question, or a dangling question is dropped so the answer stands alone.

Read a brief. Every AI window has an Upload a Brief row under the Prompt box. Click it and an upload field opens: drag the brief in or click to pick it (PDF up to 5 MB, Word, text, Markdown or images such as a photo of a page; up to 3 files). The AI reads it once, fills the options the brief states (Type, Goal, Target length, Hard max, Don't include Interviewer, Cutting style; in AI Clips the number of clips and the clip length) and writes a full Prompt: audience, message, must-include moments, structure as numbered sections, tone, what to avoid, the ending. Every filled option carries a FROM BRIEF mark and stays editable; nothing runs until you click. A field the brief does not state is left alone. The read brief is saved with the project so the other AI windows see it too; Forget brief clears it. The files themselves are never stored: they are read in your browser, sent to the AI once and gone. Its price in credits shows on the (i) next to Upload a Brief.

Two or more existing projects can be combined. You can also drag one project onto another on the dashboard; a yellow window asks you to confirm, and the dragged project stays as it is. Or: on the dashboard, open the menu (three dots) of the project that should receive the others and choose Merge with other project. Pick one or several projects in the window, choose whether the other project is kept or deleted after merging, and press Merge. The newest version of each project is used, and the merged lines land in a new version of the receiving project, so every earlier version stays exactly as it was. Each merged project’s files become files of this project and share the same stored video and audio, so nothing is uploaded or transcribed again. Picks, trims, notes and groups come along; a speaker with the same name in both projects is one person, and the Who is in this file? window offers to match the rest. A deleted project goes to the Trash, and its audio and video stay because the merged project uses them.

Reuse one transcript across projects (master transcripts)

Working on a series? You never need to import or transcribe the same recording twice. Put the project that has the transcript inside a folder: it becomes the master. Then create a New Project inside that same folder and choose Source from master transcript instead of starting fresh. The new project opens linked to that transcript and goes straight to Select, so it makes its own picks, arrangement, and versions from the same words.

Fix a word or rename a speaker in the master and every linked project updates automatically. A linked project shows a small note that its transcript lives in another project; clicking it opens the master. To hand ownership to another project, use Move master transcript in the owner card’s ⋯ menu.

Naming and the cover image

Type a project title, then click Create Project. You can also drop in a cover image (PNG, JPG, or WEBP, any size, ScriptCut auto-compresses it); it shows on the project card and on the page your client sees, a nice touch that makes a share link feel finished.

Transcribe a recording

If you do not yet have a transcript, the Transcribe shortcut on the dashboard (and the Transcribe workflow) does it for you. Upload your audio or video, and ScriptCut returns a word-level transcript with speakers detected, ready to clean up and select. Transcription minutes are part of your plan; see Settings > Transcription. ScriptCut transcribes in 54 languages: Afrikaans, Albanian, Arabic, Azerbaijani, Bosnian, Bulgarian, Catalan, Chinese, Croatian, Czech, Danish, Dutch, English, Estonian, Filipino, Finnish, French, Galician, German, Greek, Hebrew, Hindi, Hungarian, Icelandic, Indonesian, Italian, Japanese, Javanese, Kazakh, Korean, Latvian, Lithuanian, Macedonian, Malay, Maltese, Norwegian, Persian, Polish, Portuguese, Romanian, Russian, Slovak, Slovenian, Spanish, Sundanese, Swahili, Swedish, Tamil, Turkish, Ukrainian, Urdu, Uzbek, Vietnamese and Welsh. Pick the spoken language in the Language dropdown, or leave it on Auto select. Every one of them comes back word by word with timecodes, and speakers are detected in nearly all of them, so the transcript edits, selects and exports the same way in every language. AI Translate then turns any of them into 48 languages. The dropdown auto-fills from a quick background check of your recording; override it if it guesses wrong. Once transcription finishes, a Who Is Who? step lets you name each detected speaker, with a short sample of their voice you can play to tell them apart. Custom dictionary: add the names, brands and terms your recordings use and transcription spells them correctly, and Check Spelling never flags them. Set one dictionary for every project in Settings > Transcription, and add more for a folder from the folder's dashboard menu (Word list). A folder's list is used by every project inside it, including its sub-folders. A list saved earlier on a single project keeps applying to that project, but a project's menu no longer opens it. All the lists that apply are combined, up to 200 words, for new transcriptions.