Thanks — I’ll explain aligned transcript STT features and what they enable. Aligned transcripts attach start/end timestamps to words or speech chunks so each word is synchronized with the audio. Key benefits and uses: • Accurate timing: word-level or chunk-level timestamps let frontends render live captions precisely at the right moment. • Adaptive interruption handling: timestamps let the system detect overlapping speech and distinguish true interruptions (barge-ins) from backchanneling (e.g., “uh‑huh”). • Better UI features: enable highlighting current words, scrubbing to audio positions, and precise subtitle export. • Improved multimodal sync: supports forwarding transcripts aligned to TTS or other media so text and audio stay in sync. Availability and plugin notes: • All STT models in LiveKit Inference support aligned transcripts. • For STT plugins, check the

capabilities.aligned_transcript

property (Python: `if stt.capabilities.aligned_transcript:`; Node:

if (stt.capabilities.alignedTranscript)

). • Some third‑party plugins (AssemblyAI, Speechmatics, Simplismart) offer related alignment/turn-detection features; explicit word-level alignment depends on each plugin’s capabilities. If you want, tell me: • Which STT model or plugin you’re using, and • Whether you need word-level timestamps or chunk/turn-level timestamps, and I’ll give concrete setup or example payloads. Sources: Adaptive interruption handling | LiveKit Documentation | Text and transcriptions | LiveKit Documentation | AssemblyAI STT | LiveKit Documentation | Speechmatics STT plugin guide | LiveKit Documentation | Simplismart STT plugin guide | LiveKit Documentation, with interactive elements