Descript: collect recordings, then edit video like a document

Editing an interview often starts with repeatedly scrubbing through it to find the clearest sentence. Descript transcribes recordings so that users can edit around what was said. It also offers scenes, layouts and a timeline for visual work and finer adjustments. Getting started with Descript

This article follows one team workflow: collect separate recordings, organize the content and deliver a viewing link. In the teaching scenario, an editor asks a remote colleague to record equipment instructions for a short video. Recording invitations are an Enterprise team feature; text-based editing is a broader product capability. Their plan availability should not be conflated. Async recording announcement

1. Put the recording brief in the invitation. A drive is the team's workspace for projects. In a new project on an Enterprise drive, follow Record → Invite others to record → Invite to record async. Set a title, expiry and access scope, optionally include a script, then create the link. For the equipment example, a suggested brief asks three questions: when to use it, how to start and how to finish. Async recording means the two people do not need to join a meeting together. Invitation path and availability

2. Wait for processed footage before editing. When processing of the recording finishes, the inviter receives email and in-app notifications. Open the footage for project editing. The notification means the recording has been processed, not that its explanation is correct or the finished video has been published. Listen through it first to check whether the sound and content suit the instructions.

3. Organize sentences, then check the pictures. Text is linked to media: deleting a sentence removes its clip, and cutting and pasting moves the associated footage. Move “preparation before starting” ahead of the operating instructions, for example. If you remove an important reminder, restore removed media at the edit boundary. For a transcription typo alone, use transcript correction; do not assume that every text change means cutting audio. Text editing and restoration

4. Arrange the visuals for the content you kept. Divide the explanation into scenes and choose suitable layouts. Use the timeline when precise timing needs adjustment. A suggested check is to play the entire result and look for unclear references created by moving sentences: someone might say “this button” before the picture shows it. A readable transcript does not guarantee a coherent sequence of shots. Scenes and the timeline

5. Be clear about which version is published. Choose Export → Descript (web link), check access and export the share page. Later project edits do not automatically change the shared version; use Update. Downloads from the page also use the last exported version. Updating removes share-page comments but keeps project comments in the editor. If feedback needs preserving, organize the decisions before updating. Web exports and updates

Agree on a deliverable instead of a meeting time. An async invitation lets an editor describe the recording goal and access scope first, while the contributor records on their own schedule. It suits relatively self-contained explanations. An interview that needs immediate follow-up questions still benefits from synchronous discussion. The lesson is to make the expected contribution explicit, not to make every recording asynchronous.

Words define content; playback checks continuity. A transcript turns “which sentence should go?” into a visible operation while preserving its consequences for the media. For speech-led video, that can reduce the effort of locating a sentence. When a pause, shot or gesture carries meaning, watch the actual playback. A familiar editing surface should not hide what it changes.

Separate the working draft from the viewing copy. A shared link keeps the last export while the editor continues working and decides when to update what viewers see. That boundary keeps every unfinished change from immediately going out, but can also lead to “I changed it, yet others see the old version.” Delivery interfaces should distinguish the current draft, exported copy and consequences of updating—especially the loss of share-page comments.