How to Use NVIDIA Canary-1B-v2 for ASR, Translation, and Automatic SRT Subtitle Export in Python

Short summary

NVIDIA's Canary-1B-v2 enables end-to-end multilingual automatic speech recognition (ASR), real-time translation to French, German, Spanish, and Italian, and automatic SRT subtitle generation with word-level precision. The tutorial demonstrates GPU-optimized Python implementation covering 16kHz mono audio preprocessing, timestamp extraction, and batch processing workflows. Includes inference benchmarking and practical strategies for production-scale multilingual media processing.

• •Multilingual ASR with translation to four European languages using NVIDIA's latest speech model
• •GPU-optimized Python pipeline with 16kHz preprocessing and word-level timestamp precision
• •Batch processing workflows and inference benchmarking for production deployment

Generated with AI, which can make mistakes.