OpenAI’s Whisper Large v3 can now transcribe 150 minutes of audio in less than 98 seconds. The new CLI tool, ‘insanely-fast-whisper’, powered by Transformers, Optimum, and flash-attn, enables this rapid transcription. The tool is community-driven and works on NVIDIA GPUs and Macs, offering various options to maximize transcription throughput. It also supports running Whisper in different settings, making it a versatile tool for audio transcription tasks.
Read more at GitHub…