{"service":"WhisperX Batch Processing (GPU)","version":"1.0.0-gpu","platform":"NVIDIA DGX Spark (Blackwell)","endpoint":"POST /transcribe","features":["Perfect transcription (Whisper large-v3)","Word-level timestamps (Wav2Vec2 alignment)","Speaker diarization (pyannote.audio) - GPU ACCELERATED","Full GPU acceleration (Blackwell SM_121 → SM_90 spoof)","OpenAI Compatibility (/v1/audio/transcriptions)"]}