On-prem availability
Pre-recorded and streaming transcription run on-prem, as a container deployment or a virtual appliance. This page covers on-prem availability. For Speechmatics-hosted deployments, see Feature availability.
A check mark marks an available combination. An em dash means the combination is either unavailable or not established, and carries no roadmap position either way.
Region does not apply on-prem. Regions are an attribute of SaaS on Cloud, because on-prem deployments are customer-hosted.
Readiness on-prem
On-prem ships as versioned containers, so the question is which release you need rather than whether a combination is production-ready.
Feature availability on-prem depends on the container release you are running. Check Accessing images for the current releases, and the release notes for the change that introduced a given feature.
Pre-recorded transcription features on-prem
Pre-recorded transcription runs on the Standard, Enhanced, and Melia 1 models on-prem.
Language coverage and selection
Output and formatting
Transcript content and tagging
Speakers and channels
Audio input
Audio events on-prem requires a GPU container.
Operational
Add-ons
Translation on-prem runs in a separate inference container. See Translation GPU inference container.
Streaming transcription features on-prem
Streaming transcription runs on the Standard and Enhanced models on-prem.