WAAS — Whisper as a Service — is an open-source project from Schibsted with 2,000+ stars on GitHub that wraps OpenAI’s Whisper speech-to-text model in a self-hosted service with a graphical interface, an API and a job queue.
Whisper is excellent at transcription, but running it well is an operations problem: GPU time is precious, long files take minutes, and several users submitting at once will trample each other. WAAS solves the serving layer. You deploy it on your own infrastructure, users submit audio through a web GUI or programmatically through the API, and a queue schedules transcription jobs against the Whisper model in an orderly fashion, returning text when each job completes. That makes it a pragmatic fit for newsrooms and media companies — Schibsted is a large media group, and the project reflects that origin — as well as any organisation that wants Whisper’s accuracy without sending audio to a third-party cloud or paying per-minute API prices. It is Apache-2.0 licensed and still actively maintained, with updates in 2026, which is more than can be said for many Whisper wrappers from the same era.
WAAS is free and open source under the Apache-2.0 license. You pay only for the servers and GPU time it runs on — typically far cheaper per hour of audio than per-minute transcription APIs at volume.
Media companies, researchers and internal-tools teams who transcribe a lot of audio and want Whisper-quality speech-to-text on their own infrastructure, with a queue and an API instead of ad-hoc scripts.