StableStream™ is LansonAI's core technology for continuous, stable, readable live text in Lanson Live. It describes output behavior, not a model you switch to.
Real-time speech recognition produces changing information.
A naive interface can expose those changes directly:
we should probably
we should probably ship
we should probably ship the
we should probably ship this
we should probably ship this next
we should probably ship this next weekIf earlier words are repeatedly replaced or rearranged, the interface may technically be updating quickly while becoming difficult to read.
StableStream™ addresses this boundary between model updates and human-readable output.
Ready to read, not racing to display
The fastest possible token is not always the most useful token.
A live interface must balance:
- response latency
- linguistic uncertainty
- correction quality
- visual stability
- reading continuity
StableStream™ is designed around that balance.
Its goal is to allow new information to arrive continuously without making previously presented information unnecessarily unstable.
Recognition stability and visual stability
These are related but different concepts.
Recognition stability describes how confident the system is that speech has been interpreted correctly.
Visual stability describes how much already-presented content moves or changes on screen.
A good live speech experience needs both.
StableStream™ operates at this boundary.
Application behavior
Applications consuming a live stream should distinguish between content that is still evolving and content that has become stable.
The API documentation describes the exact events and state transitions available to clients.
This page describes the underlying principle:
Live text should evolve without forcing the reader to repeatedly reconstruct what they have already understood.
What the public Realtime API delivers
On WS /v1/audio/transcriptions/stream, each successfully processed utterance produces one conversation.item.input_audio_transcription.completed event. Its transcript.state is stable and its transcript.revision is currently 0.
The endpoint does not stream partial text or later revisions of the same utterance. Optional refinement runs before the completed result is emitted. This lets your UI display completed segments without repeatedly replacing earlier words; it does not guarantee recognition accuracy or ordering between concurrently processed segments.
Check transcript.safe_for_downstream before triggering actions. A true value is a basic readiness signal, not proof that the transcript is correct. Order text by utterance_index and handle errors and language uncertainty explicitly.
Relationship to the engine and processing mechanism
- Trinity Engine coordinates speech processing and context.
- LiveFinal™ describes progressive correction and segment finalization.
- StableStream™ describes how the resulting live text stays stable and readable.
Use the Transcript Lifecycle and Server Events pages as the wire contract. Do not infer extra event types from a technology name.