Zero-shot latent-inpainting model that regenerates speech spans to match the adjacent voice, delivery and room tone. It powers word-level audio editing in Descript.
Continuous neural codec plus flow-matching transformer; masked audio latents generated in parallel, guided separately by transcript and surrounding audioSource for Architecture (opens in a new tab)
No standalone model API ID is published for this editing feature.
Access
Highlight recorded words and choose Regenerate. Replacing words requires the speaker’s authorized custom AI speaker; smoothing unchanged words uses the existing recording.
Regions
Managed service; exact model-processing region not verified
Current Regenerate editor workflow. Do not inherit the multilingual ElevenLabs TTS language list.
Input requirements
No stock voices for changing recorded speech. Selections cannot cross multiple edit points, scene boundaries or ignored text. Keep an already regenerated clip before editing it again.Source for Input requirements (opens in a new tab)
Stock-voice exclusion is independently documented in the change-meaning guide.
Billing basis
Manual effects scale with job size. Underlord charges separately for its agent reasoning and the tools it invokes; identical prompts can use different credit amounts.Source for Billing basis (opens in a new tab)
Included plan allowances
Hobbyist: 400 AI credits + 10 media hours per editor/month, Creator: 800 AI credits + 30 media hours per editor/month, Business: 1,500 AI credits + 40 media hours per editor/month, Free: 100 one-time AI credits + 1 media hour/month; limited feature trialsSource for Included plan allowances (opens in a new tab)
Editor seats contribute to a shared Drive pool. Enterprise has custom allowances. Plan limits and export resolution are not model context/output limits.
Prices retain their published currencies and billing units below.
Published billing units · original currency · observed 13 Oct 2026
Category
Usage
Rate
Unit
Subscription
Hobbyist · monthly billing1
$24.00
editor / month
Subscription
Hobbyist · annual billing2
$16.00
editor / month, billed annually
Subscription
Creator · monthly billing1
$35.00
editor / month
Subscription
Creator · annual billing2
$24.00
editor / month, billed annually
Subscription
Business · monthly billing1
$65.00
editor / month
Subscription
Business · annual billing2
$50.00
editor / month, billed annually
Credit bundle
Optional shared credit top up3
$35.00
350 AI-credit bundle
Credit bundle
Optional shared credit top up3
$80.00
1,000 AI-credit bundle
Credit bundle
Optional shared credit top up3
$200.00
4,000 AI-credit bundle
Rate conditions
Shared editor subscription including credit/media allowances and other features; AI operations consume credits. Not a flat per-model or per-question price.
Monthly-equivalent shared subscription with annual commitment. Included monthly credits do not roll over. Not an independent inference tariff.
Creator/Business only. Shared across the Drive, expires 12 months after purchase, consumed after monthly allowance. Credit cost of an operation varies; do not convert this into a universal media-minute rate.
Descript help states that production models use no Descript user data; internal research/development uses data only from users who opted in. Enterprise data sharing is disabled by default. Partner agreements prohibit using submitted data to train partner AI. This is not a blanket zero-retention promise; project storage and deletion have separate policies.
Supply exactly one of project_id or project_name; optional composition_id requires project_id. Callback URL supports completion/failure notifications.Source for Request limits (opens in a new tab)
Model routing
model selects the editing agent’s LLM, not this media model. GET /agent/models is authenticated and dynamic; documentation examples are not the full available catalog.Source for Model routing (opens in a new tab)
No additional API access fee for paying users; operations use the account’s AI credits and media minutes. Agent reasoning and the tool are both metered.Source for Billing basis (opens in a new tab)
Endpoint pricing has not been collected. Check the provider for applicable rates.
Data & enterprise terms
Descript help states that production models use no Descript user data; internal research/development uses data only from users who opted in. Enterprise data sharing is disabled by default. Partner agreements prohibit using submitted data to train partner AI. This is not a blanket zero-retention promise; project storage and deletion have separate policies.
Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.
01 / INITIAL INPUTProcess & retain
The system processes eligible input and retains reusable state.
02 / LATER REQUESTCheck available state
Reuse depends on matching input, session state and the cache policy.
03 / CACHE HITReuse & generate
Reusing state can reduce repeated work. Any charges follow the billing rules below.
Descript editor
Not verified
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
How it works
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
Minimum reusable input
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
Lifetime & refresh
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
How to enable it
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
Usage fields
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
Billing rules
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
Limits & data boundaries
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Descript Agent API
Not verified
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
How it works
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
Minimum reusable input
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
Lifetime & refresh
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
How to enable it
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
Usage fields
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
Billing rules
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
Limits & data boundaries
Not collected
No model-specific discounted prompt-cache contract collected. Project reuse, stored voices and persistent editing conversations do not establish caching.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
What the measurements say
No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.
Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.
The record behind the model
Why it is included
Descript identifies its media-editing models as proprietary and reserves service/model rights. The record is a documented proprietary model generation or explicitly labeled managed enhancement family; project-page code is not released model weights. Separately downloadable DAC and Mouse research weights are excluded.
