FFmpeg 8.0 adds native Whisper transcription filter and cross-vendor AV1 GPU encoder

FFmpeg 8.0, codenamed 'Huffman' and released in late August 2025, introduces two significant pipeline features: a native Whisper audio filter for speech-to-text transcription and av1_vulkan, a cross-vendor GPU-based AV1 encoder. The Whisper filter, powered by whisper.cpp, allows developers to generate SRT or JSON captions directly within an FFmpeg command without relying on separate transcription services. The av1_vulkan encoder leverages Vulkan video-encode extensions, enabling GPU-accelerated AV1 encoding across multiple hardware vendors rather than being tied to NVIDIA or Intel-specific solutions. Both features require FFmpeg 8.0 or later, as they are absent from earlier versions, with version 8.0.1 recommended as a more stable build. Developers whose Linux distributions still ship older FFmpeg versions are advised to use a static build or an updated package repository to access these capabilities.