But sadly, WebAssembly cannot directly use these x86 assembly code as it is not compatible or certain instructions are not supported. x264), they use assembly language like x86 to speed up the process.

Leverage CPU capabilitiesįor most of libraries work with ffmpeg (ex. It is a normal situation as web workers are simulating threads in FFmpeg and it is good as we don’t want ffmpeg.wasm to block our main thread. When you run n(), you might find tons of web workers spawned. But in ffmpeg.wasm, to enable pthread / mutli-threading support to speed up, it is a required data type to use. SharedArrayBuffer is a pretty new data type in JavaScript, right now most of the browsers still lack of fully support due to security issues. ffmpeg.FS('readdir', '/') : list files inside specific pathįor full list of APIs, you can check: SharedArrayBuffer.ffmpeg.FS('unlink', 'filename') : delete a file in MEMFS.

ffmpeg.FS('readFile', 'filename') : read file from MEMFS as output of ffmpeg.wasm.

ffmpeg.FS('writeFile', 'filename', data) : write file to MEMFS as input of ffmpeg.wasm.

It is essential as we want to keep minimal changes to FFmpeg source code and reserve its maxium similarity to original command line interface.

(In ffmpeg.wasm we use MEMFS / Memory File System) You can imagine this FS just like a hard disk where you can put the input file and pull the output file from ffmpeg.wasm command. When you check the API of ffmpeg.wasm, you may find an API called ffmpeg.FS() to handle File System operations. Download ffmpeg-core.js from remote server (which is around 25MB)ĭepending on your network speed and host machine hardware, this operation can take up to minutes to complete.To fully utilize/understand the power of ffmpeg.wasm (also for most WebAssembly libraries), few technical details you might want to know: Inside ffmpeg.load()įfmpeg.load() is a required API to call before you call others. A demo to transcode AVI to x264 MP4 Behind the scene

Ffmpeg javascript example