ffrwd-package

Speech into the pipeline: whisper transcribes or translates an audio stream and the words leave as cues, hosted in wasm with inference on the host.

RF-DETR detection and instance segmentation, hosted in wasm.

QR codes in video: the Rust counterpart to ffrwd/jsqr, same algorithm, for timing

Voice activity detection for ffrwd: Silero VAD in wasm, the audio passed through untouched and every span of speech leaving as a cue.

ffrwd/yolo26: YOLO26 detection and segmentation hosted in wasm; detect makes rows, segment_mask makes a matte, and ffmpeg does the rest. AGPL-3.0.

ffrwd/faceage: ages on faces, for blurring the children in a video and nobody else

QR codes in video: a JavaScript ffrwd module, componentized to wasm

ffrwd/longpipe: virtual-background matting for ffrwd pipelines, weights from sb2702/longpipe (MIT).

PyroWave, the GPU wavelet codec, as ffrwd codec modules over wasi:webgpu

Video intelligence for ffrwd: describers that write what a video shows and says as metadata tracks, and the embeddings a query searches

The official ffrwd packages: Whisper subtitles, RF-DETR and YOLO detection, face age, VAD, HLS/DASH ladders and more, free to install.

ffrwd/moq: Media over QUIC in both directions, hosted in wasm; publish is a COPY destination and subscribe a FROM relation.

ffrwd/index: embedding vectors woven into a video stream and read back out of it. A format, a codec, a command line and the ffrwd package