Original Lyrebird/Descript speech-inpainting generation for replacing or adding words while matching the surrounding recording. Current access to the original weights is not established.
Historical text-informed inpainting research: bidirectional/style-conditioned autoregressive models and a duration-predictor alternativeSource for Architecture (opens in a new tab)
The article describes multiple explored architectures, not a public checkpoint selector for each.
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 · historical
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.
