AD
Descript logo

Descript

Freemium🇺🇸Near-Breakeven

Edit video and audio like a doc — then let AI finish the rest

91

Overall score

45

Heat score

Pricing

Free$0/month
Hobbyist$24/month (or $16/month billed annually)
Creator$35/month (or $24/month billed annually)
Business$65/month (or $50/month billed annually)
EnterpriseCustom (contact sales)

Technical Specs

Inputs

Audio File, Video File, Screen Recording, Webcam Recording, Remote Interview, Existing Transcript, Zoom Recording

Outputs

Edited Video, Podcast Episode, MP4 Export, MP3 Audio, Transcript, SRT Subtitles, Social Media Clip, AI Generated Video, Blog Post From Transcript, GIF

AI Type

Speech

Model Architecture

Transformer

Daily Prompts

N/A

Context Length

N/A

Output Quality

Accuracy

91%

Content

90%

Reasoning

88%

Company Profile

Company

Descript Inc.

Founded

2017

HQ

San Francisco, California, USA

Employees

187

Total Raised / Total Funding

$101M

Revenue

$55M

Valuation

$550M

ARR

$55M

CEO

Laura Burkhauser

Overview

Estimated Paid Users

500K

Current estimate

Total Earnings Till Date

$55M

+5.60% from last month

Market Share

3.8%

Current share

Average Session

55

Per active user

Hallucination Rate

5%

Model quality signal

Growth Rate

+4.93%

Monthly active users

Burn Rate

N/A

Total expenses / years active

Paid User Gain

+20.59%

Monthly paid user trend

No demo video available yet.

Platforms

AD

Profit Analysis

$75M

Total Loss

$14.3M

Total Profit

$128M

Performance Metrics

Accuracy

91%

Context

90%

Reasoning

88%

Safety

90%

Benchmarks

No benchmark scores available.

Descript Models

Descript Transcription Engine v1

Type: STT

Description: Initial proprietary speech-to-text transcription model powering the core Descript editor. Combination of automated ASR (automatic speech recognition) and human correction layers. Targeted podcast and audio journalism workflows with multi-speaker support and high accuracy on professional-quality recordings. Text-edit-to-audio-edit binding was the platform's foundational technical innovation — edit the transcript, the audio changes accordingly.

Architecture: Transformer

Overdub v1 (Voice Clone)

Type: Audio

Description: Descript's first-generation proprietary voice cloning model. Trained on 10+ minutes of user-recorded speech. Could regenerate words and short phrases in the user's voice for in-place corrections. Required 24-hour training and authorization time. Raised significant public attention around AI voice authenticity — Descript required verbal consent and passphrase to train, prohibiting non-consensual voice cloning. Generated controversy alongside excitement.

Architecture: Transformer

Studio Sound (Audio Enhancement)

Type: Audio

Description: Proprietary one-click audio enhancement model that removes background noise, room echo, reverb, and frequency irregularities from any recording environment to produce studio-quality sound. Runs on Descript's own AI audio stack — not a licensed third-party tool. Consistently cited as the most productivity-saving Descript feature. Available as a standalone toggle and as an Underlord AI Action. Processing time: under 10 seconds per audio segment.

Architecture: Transformer

AI Eye Contact

Type: Vision

Description: Proprietary computer vision model that corrects gaze direction in recorded video — making speakers appear to look directly into the camera even when reading from a script or notes. Uses face landmark detection and neural rendering to redirect eye gaze frame-by-frame. Particularly valuable for interview and educational content where direct eye contact builds audience trust. Available in Creator and Business plans.

Architecture: Transformer

Overdub v2 (Rebuilt Voice Synthesis)

Type: Audio

Description: Complete rebuild of Overdub with significant quality improvements. Added text-to-speech capability for generating complete voiceovers from typed scripts without recording. Near-instant voice authorization (from 24 hours to minutes). Multi-speaker Overdub for correcting other voices in a project with permission. Improved naturalness and reduced synthetic artifacts especially on longer passages. Current voice cloning model as of March 2026.

Architecture: Transformer

Funding Rounds & Investors

Total Funding

$101M

Rounds

4

Funding Update

$50M

Nov 2022

Raised $50M Series D led by the OpenAI Startup Fund at approximately $550M valuation — Descript was the second company to receive funding from the OpenAI Startup Fund after Mem. Announced November 15, 2022 alongside a major product redesign including Storyboard, Write Mode, and AI background removal. OpenAI COO Brad Lightcap joined as a key backer. Total raised: $100M. a16z, Redpoint, and Spark Capital all continued. Daniel Gross (ex-YC partner, builder of Cue and Pioneer) invested as an angel. The round came as Andrew Mason described being on the 'verge of a generational change in the way we create content — fueled by AI.' At the time of the round, Descript had not disclosed ARR publicly; by August 2025, confirmed ARR was $55M growing at 75% YoY.

Funding Update

$30M

Jan 2021

Raised $30M Series C led by a16z at approximately $260M valuation — announced January 12, 2021. Total raised: $50M. COVID-19 accelerated demand as remote work forced companies into video communication and content creators surged. Descript was positioned as 'the Google Docs moment for video editing.' The round funded expansion into video-first features, remote recording (Rooms), and AI capabilities including Overdub voice cloning and early AI transcription improvements.

Funding Update

$15M

Mar 2019

Raised $15M Series B led by Redpoint Ventures with continued participation from a16z and new investor Spark Capital. By this point Descript had expanded from audio-only to video editing with the same text-based paradigm. NPR, VICE, Pushkin Industries, The Washington Post, and The New York Times were among early enterprise adopters. The video expansion was driven by observation that most popular podcasts were simultaneously producing YouTube video versions — a trend requiring two separate editing workflows that Descript could unify.

Founders/Team

AM

Andrew Mason

Founder & Executive Chairman

LB

Laura Burkhauser

CEO

Direct competitors

No direct competitors available.

Change Log / Major Updates

2024 · Mar 1

Rebuilt Overdub from scratch: significantly improved voice quality, text-to-speech capability (generate voiceovers from scratch without recording), near-instant voice authorization (from 24 hours to minutes), and multi-speaker Overdub. Described by Descript as: 'We basically took our beat-up old AI into the shop, rebuilt the engine, and outfitted it with a set of all-new space-age capabilities.' Enabled complete voiceover creation without recording a single word. Also launched Scenes and Layouts for structured video composition.

2025 · Apr 1

Released the Season 6 redesign — described by Descript as 'the biggest improvements we've ever made' in preparation for the Underlord agent launch. New layout with cleaner timeline management, improved layer handling, and a rebuilt web app exiting beta. Set the architectural foundation for Underlord to operate across all features as a unified AI agent rather than a collection of discrete AI buttons.

2025 · Jul 1

Launched Underlord — Descript's AI editing agent built across the entire platform. Pre-packaged AI Actions: Edit for Clarity, Remove Filler Words, Remove Retakes, Add Chapters, Generate Show Notes, Find Shareable Clips, Write YouTube Description. Supports natural language instructions for complex multi-step editing tasks. Model picker added in subsequent update supporting Claude Sonnet 4.5, Claude Sonnet 4.6, and Gemini 3.1 Pro. CEO Andrew Mason stepped down; Laura Burkhauser became CEO on August 29, 2025. $55M ARR confirmed, growing at 75% YoY, accelerating.

Compliance, Integrations & Support

Industry: Not specified

Compliances: Not specified

Integrations: YouTube, Spotify, Apple Podcasts, Zoom, Zapier, Slack, SquadCast, OpenAI, Anthropic Claude Haiku, Descript API

Support:email, help center, community forum, priority support, enterprise support

Target audience: Podcasters, Content Creators, Video Editors, Marketers, L&D Teams, Sales Teams, YouTubers, Journalists, Educators, Businesses

Supported languages: English, French, German, Spanish, Portuguese, Italian, Japanese, Korean, Chinese, Dutch, Polish, Danish, Swedish, Norwegian, Finnish, Turkish, Czech, Romanian, Hindi, Catalan, Croatian, Hungarian, Latvian, Lithuanian, Malay, Slovak, Slovenian

Descript Acquisitions

No acquisition records available.

AD

Reviews & Rating

0 reviews

No reviews yet

Be the first to share how Descript performs for your workflow.

0.0

Accuracy

0.0

Ease of Use

0.0

Output Quality

0.0

Security

0.0

More About Descript

There is a certain kind of product insight that, once you hear it, makes you wonder how the entire industry missed it for so long. Andrew Mason had his while building Detour — an audio tour app that Bose eventually acquired in 2017. He was constantly working with spoken audio, constantly fighting audio editing tools built for music producers, constantly frustrated that fixing a mistake in a recording meant hunting for the exact waveform instead of just... fixing the words.

The insight: audio is just a transcript with sound attached. Edit the transcript, and the audio edits itself. He built Descript around that idea in 2017. Video followed. Underlord — Descript's AI co-editor — followed that. And then, in mid-2025, Andrew stepped back to Executive Chairman, handed the CEO seat to Laura Burkhauser, and watched the company he'd built hit $55M ARR growing 75% year-over-year.

How the Editing Actually Works

Import any video or audio into Descript and you get three things immediately: a transcript, a timeline, and a canvas. From that point, the product inverts every assumption traditional editing makes.

  • Delete words from the transcript → those words disappear from the video
  • Highlight a sentence and hit "remove filler words" → every "um," "uh," and repeated phrase across the entire recording vanishes in seconds
  • Don't like how you look? Eye Contact corrects your gaze to face the camera directly even if you were reading from a script
  • Bad room acoustics? Studio Sound applies broadcast-quality noise removal with one click
  • Flubbed a line after recording? Regenerate rewrites the audio in your own voice cloned voice without re-recording

These are not features bolted onto a conventional timeline editor. They are what Descript is built around. The conventional editing tools — razor cuts, waveform manipulation, J-cuts — are available, but nobody who uses Descript regularly reaches for them first.

Underlord: The AI Director in the Room

Underlord is Descript's AI co-editor, and it is ambitious in a way that most editing tools are not. It doesn't just help you clean up a recording — it helps you figure out what the recording should be. Give it a 90-minute interview, and Underlord can identify the best 10 minutes, write a YouTube description, generate timestamped chapters, create three social clips with captions, and draft a show notes blog post. All before you've made a single manual cut.

The September 2025 relaunch restructured Underlord around an AI credits model — a single pool that covers every AI action, from a Studio Sound enhancement (10 credits/use) to a full video translation with dubbing (15 credits/minute). The goal: replace 17 different feature-specific limits with one flexible budget. The Creator plan's 800 monthly credits handles most weekly creators without hitting the ceiling. Top-up packs let power users buy more credits without upgrading their entire plan.

Descript Rooms: Remote Recording Done Right

In 2023, Descript acquired SquadCast — the remote recording platform used by professional podcasters for its separate audio/video track recording and studio-quality output. That technology became Descript Rooms: a remote recording studio built directly into Descript that records each participant on separate tracks, up to 4K resolution, with local recording that uploads in the background so dropped connections don't ruin the session. The result is a single tool that covers the entire production workflow: schedule and record → transcribe and edit → enhance and package → publish and distribute.

The Business Reality

$55M ARR. 75% year-over-year growth accelerating. $101M raised with OpenAI leading the $50M Series C at a $550M valuation — alongside a16z, Spark Capital (whose GP Nabeel Hyatt has been a Descript board member and champion since the earliest rounds), and Redpoint Ventures. Individual investors include Casey Neistat, Tobi Lütke (Shopify), Naval Ravikant, Shishir Mehrotra (Coda/Superhuman), Lenny Rachitsky, and Rahul Vohra (Superhuman) — a roster that reads like a who's-who of people who understand what it means to build a product that makes creators genuinely better at their craft.

Descript FAQ's

What is Descript and what makes it different from Adobe Premiere or Final Cut Pro?

Descript is an AI-powered audio and video editor built on a radically different paradigm: instead of manipulating a waveform or timeline, you edit the automatically generated text transcript and the audio/video changes underneath. Delete a sentence from the transcript and it disappears from the media. Rearrange paragraphs and the clips rearrange. This makes editing accessible to anyone who can use a word processor — no professional editing skills required. On top of this foundation, Descript adds AI features including Overdub (voice cloning for corrections), Studio Sound (one-click noise removal), Underlord (AI editing agent), Rooms (remote recording studio), AI Eye Contact, and AI Green Screen.

What is Underlord and what can it do?

Underlord is Descript's AI editing agent, launched in July 2025 with Season 6. It's an all-purpose assistant embedded across every part of Descript's workflow. Pre-packaged AI Actions — Edit for Clarity, Remove Filler Words, Remove Retakes, Add Chapters, Generate Show Notes, Find Shareable Clips, Write YouTube Description — execute specific editing jobs with one click. The underlying agent supports natural language instructions ('clean up this interview and find the three best moments') and uses Claude Sonnet 4.5/4.6 or Gemini 3.1 Pro (selectable via model picker) as its reasoning engine. Underlord links its edits back to specific transcript locations so you can review what changed.

What is Overdub and how accurate is it?

Overdub is Descript's proprietary voice cloning feature — record 10+ minutes of your own voice and Descript trains a model that can regenerate speech in your voice from typed text. The primary use case is correcting mistakes: instead of re-recording an entire paragraph because you mispronounced one word, type the correction and Overdub generates it in your voice seamlessly. Descript claims 95% accuracy; independent testing typically finds 88–98% accuracy depending on audio quality and vocabulary. A 2025 model rebuild significantly improved quality, added text-to-speech capability, and reduced authorization time from 24 hours to near-instant. Overdub works best for 1–10 word corrections; longer passages can sound slightly synthetic.

What is Studio Sound?

Studio Sound is Descript's one-click AI audio enhancement feature. Upload or record audio in any environment — home office, coffee shop, outdoors — and Studio Sound removes background noise, echo, and reverb while equalizing frequency response to sound like a professional recording studio. Typically reduces audio cleanup from 30–60 minutes of manual EQ and noise-gate work to under 10 seconds. Consistently cited by Descript users as its most productivity-saving feature. It runs as an Underlord AI Action and as a standalone toggle in the audio editor.

How much does Descript cost?

Descript offers four pricing tiers. Free: basic editing, 1 hour transcription per month, limited Underlord actions, 720p export. Hobbyist: $12/month, 10 hours transcription, Overdub, Studio Sound, 1080p export. Creator: $24/month, unlimited transcription, unlimited Underlord actions, 4K export, multi-speaker Overdub, AI Green Screen, Rooms. Business: $40/user/month, everything in Creator plus SSO, shared team libraries, custom branding, and enterprise support. Annual billing saves approximately 33% on all plans.