The transcription stack that runs where your data has to stay.
Makimoto Kawa is an MIT-licensed transcription pipeline with managed APIs for real-time and post-call workloads. Hosted in Singapore from 1 July 2026, with country-specific deployments and a self-hostable release to follow.
What is Kawa?
Kawa is a transcription pipeline with two managed APIs at launch: one for real-time use cases like live captioning and voice agents, one for post-call workloads like recorded calls and batch analytics. Both run the same five-stage pipeline: audio resampling, voice activity detection with speaker diarisation, noise filtering, speech-to-text, and post-processing. The orchestration layer is open source. Each component is replaceable.
Why Makimoto?
A pipeline you can actually read
The Kawa orchestration layer is published on GitHub under the MIT licence at launch (01 July 2026). Read the source, file issues, fork it, contribute back. Internal pipeline components open progressively through 2026 and 2027, with no source-available carve-outs and no Business Source License surprises.
Your data stays where you deploy
Customer audio and transcripts are processed in country, not just stored there. Kawa's first release runs exclusively in Singapore. Country-specific deployments and a self-hostable container release follow, for jurisdictions where data residency is non-negotiable.
Modular by design
Voice activity detection, diarisation, noise handling, the speech-to-text model, post-processing — each stage is replaceable. Tune for your language, your domain, or your latency budget. Composability is the architecture, not a feature.
Kawa is live
Kawa's post-conversation transcription API is live as of 1 July 2026. Built for recorded calls, voicemail, and large-scale analysis of archived customer interactions: submit an audio recording, retrieve a completed transcript with speaker separation and segment-level timestamps.
Each account includes a free monthly allowance of 1,000 minutes. The playground runs the same endpoints as production, so what you test is what you ship. A real-time transcription API for live voice agents and captioning follows as a subsequent release.
Try it now