
A tutorial demonstrates building a speech recognition and translation pipeline using NVIDIA Canary-1B-v2 in Python. The workflow covers audio preparation, English ASR, multilingual translation, timestamp generation, and SRT subtitle export. The pipeline supports long-form transcription, batch processing, and inference benchmarking on GPU-enabled runtimes.
Tap to vote and see what everyone thinks.
Summary by ByteBrief
Canonical reveals Myna, its local speech-to-text app