STT: ElevenLabs STTv2 (Scribe v2 Realtime) support

[varghesepaul]

Summary

Adds support for ElevenLabs Scribe v2 Realtime streaming STT with ~150ms latency.

Features

• WebSocket-based streaming transcription
• Configurable commit strategies (VAD/manual)
• Word-level timestamp support
• Automatic reconnection handling
• Comprehensive error handling

API Options

• model_id: Model selection (default: scribe_v2_realtime)
• language_code: Language support (optional)
• commit_strategy: "vad" (default) or "manual"
• include_timestamps: Enable word-level timestamps
• VAD parameters: threshold, silence duration, speech duration

Implementation Details

• Follows Deepgram STTv2 pattern for consistency
• Uses RecognizeStream base class (modern API)
• Proper usage tracking via RECOGNITION_USAGE events
• Session override support via update_options()

Known Issues

ElevenLabs API currently returns duplicate transcripts in some scenarios. I've reported this to ElevenLabs (https://github.com/elevenlabs/elevenlabs-python/issues/686). No explicit deduplication logic added as it risks removing valid repeated content.

Documentation

STT - Realtime : https://elevenlabs.io/docs/cookbooks/speech-to-text/streaming , https://elevenlabs.io/docs/api-reference/speech-to-text/v-1-speech-to-text-realtime