Descript
by Descript, Inc.
Text-based audio & video editing, transcription, and AI media tools
Features
- Text-based editing: edit audio/video by editing the transcript (delete or rearrange text to cut/trim media).
- Automatic transcription (multi-language) with speaker detection and editable transcripts.
- Studio Sound: one-click background-noise removal and voice clarification.
- AI speech (formerly marketed as Overdub): text-to-speech from a script with stock voices or an authorised custom “AI Speaker” voice clone of your own voice (Descript help center, 2026-10-02; the “Overdub” name no longer appears in the help index).
- Underlord: Descript’s AI co-editor/video editor that automates editing tasks (per descript.com, from the Hobbyist plan up); other AI tools include eye-contact correction, retake removal and video regeneration (descript.com).
- Remove filler words and silence automatically; smart cleanup tools.
- Video editing features: multi-track timeline, captions, overlays, multiformat exports (widescreen, square, vertical).
- Automatic captions / subtitles export (SRT, VTT) and closed caption support.
- Automatic clip creation / highlight reels and social-media friendly repurposing tools.
- Screen recording, multitrack recording, and remote interview recording (Stream & Call features).
- AI Assistants & templates: AI-generated summaries, show notes, and AI templates for clip creation.
- Integrations: export/roundtrip with Final Cut Pro, Adobe Premiere, publishing integrations and podcast hosts.
Superpowers
Descript’s distinguishing feature is its text-first editing approach: media is edited by editing a transcript, so ums, pauses and whole sections can be removed by deleting text (opinion: this may suit creators who find timeline editors slow). The platform bundles transcription, audio cleanup and multi-format export.
Who this is for
- Podcasters who want fast transcription, filler-word removal, and easy chaptering.
- YouTubers and short-form video creators repurposing long-form footage into clips and reels.
- Marketing teams producing explainer videos, product demos, and social content quickly.
- Educators and trainers creating captioned course material and screencasts.
What you gain
- Editing via text instead of scrubbing timelines (time savings not measured here).
- One-click audio cleanup (Studio Sound).
- Automatic reformatting and captions for multi-platform publishing.
Plans (descript.com/pricing, read 2026-10-02)
| Plan | Includes |
|---|---|
| Free | 60 media minutes/month, 100 one-time AI credits, 720p export with watermark |
| Hobbyist | 10 media hours, 400 AI credits/month, 1080p without watermark, Underlord |
| Creator | 30 media hours, 800 AI credits/month, 4K, Underlord with 20+ AI tools, royalty-free stock media |
| Business | 40 media hours, 1,500 AI credits/month, Brand Studio, dubbing in 30+ languages, custom avatars, SLA support |
| Enterprise | SSO/SCIM, custom legal terms, AI controls |
| All paid plans: unlimited projects, transcription in 25 languages, live collaboration. Current prices: see the vendor pricing page https://www.descript.com/pricing |
Quick practical examples
- Turn a 40-minute podcast into a 3–5 minute social clip pack: transcribe, use “Create Clips” to surface highlights, then export vertical and square versions with captions.
- Fix a recorded line: generate a replacement sentence with your custom AI Speaker voice (requires setting up and authorising the voice).
- Produce accessible training videos: record screen + microphone, auto-generate captions (VTT/SRT), and publish with embedded captions.
Integrations & ecosystem
- Exports to common caption formats (SRT, VTT), and accepts media for roundtrip editing in Final Cut Pro / Premiere via exports.
- Podcast hosting and distribution integrations (major hosts), Slack and other productivity tools for team workflows.
Limitations & considerations
- Credit/quota model: AI features and media hours are metered by plan (see the table) or plan quotas — heavy users should size their plan accordingly.
- Edge cases in transcription: automatic transcripts generally need proofreading for accuracy, speaker labels, and specialized vocabulary.
- Voice cloning ethics & consent: a custom AI Speaker requires you to authorise your own voice (help center: “authorizing a custom AI Speaker that speaks in your own voice”). Use responsibly.
- For projects that need extremely fine-grained frame-by-frame video compositing or VFX, traditional NLEs (Premiere, Final Cut, DaVinci Resolve) may still be preferable.
Company and security
Per descript.com: SOC 2 Type II, SAML SSO/SCIM, GDPR compliance (vendor claims); enterprise customers named include Amazon, Canva, Salesforce, Spotify, Reuters, CBS, Microsoft and The New York Times. The about page says Descript has raised USD 100M in total, primarily from the OpenAI Startup Fund, Andreessen Horowitz, Redpoint and Spark Capital (vendor claim). Founder: Andrew Mason (previously founded Detour, acquired by Bose in April 2018). Reported milestones (TechCrunch tag page, read 2026-10-05): seed round 2017, 30M raise in 2021, OpenAI investment in 2022.
Related
ai-slides folder siblings for video: gling, opus-clip, remotion. Slide tools: ai-presentation-tools-overview.
Open items
- HQ, current ownership, user numbers and any 2026 company news not verified (Wikipedia page returned 404; web search budget exhausted).
- Whether the “Overdub” brand is fully retired: only the help-center index (no Overdub entries) was checked.
- Folder placement: this note sits in
ai-tooling/; it is a video/audio editor, not a slide tool, so it is a sibling of the video notes inai-slides/only by theme.
Sources (accessed 2026-10-02)
- https://www.descript.com
- https://www.descript.com/pricing
- https://www.descript.com/about
- https://help.descript.com/llms.txt
- https://techcrunch.com/tag/descript/ (accessed 2026-10-05; founder and funding history, secondary press index)