fix(silero): Persist RNN state between VAD inference calls

[toubatbrian]

Fixes two bugs in the Silero VAD ONNX model wrapper that caused the RNN state to never be updated between inference calls:

1. Wrong output key: The ONNX model outputs the state as stateN, not state. The state was being read from the wrong key and never stored.

2. Wrong context slice: Audio context was taken from the beginning of the input buffer instead of the end, breaking audio continuity between chunks.

Before

• RNN state remained at initial zeros forever
• Each inference ran without temporal context
• VAD accuracy degraded, especially for speech boundary detection

After

• RNN state properly persists across inference calls
• Model maintains temporal memory for accurate speech detection

Mirrors fix from Python SDK: livekit/agents#4437

[changeset-bot]

🦋 Changeset detected

Latest commit: 85009db

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 17 packages

Name	Type
@livekit/agents-plugin-silero	Patch
@livekit/agents-plugin-baseten	Patch
@livekit/agents-plugin-deepgram	Patch
@livekit/agents-plugin-openai	Patch
@livekit/agents-plugin-rime	Patch
@livekit/agents	Patch
@livekit/agents-plugin-anam	Patch
@livekit/agents-plugin-bey	Patch
@livekit/agents-plugin-cartesia	Patch
@livekit/agents-plugin-elevenlabs	Patch
@livekit/agents-plugin-google	Patch
@livekit/agents-plugin-inworld	Patch
@livekit/agents-plugin-livekit	Patch
@livekit/agents-plugin-neuphonic	Patch
@livekit/agents-plugin-resemble	Patch
@livekit/agents-plugin-xai	Patch
@livekit/agents-plugins-test	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR