FFmpeg

FFmpeg for Audio is the FFmpeg build we use at Fremen — compiled with a curated set of audio codecs that standard packages omit due to licensing and build complexity. This is what runs under every Fremen product, and we make it available as pre-compiled static binaries for Linux x86-64, Linux ARM64, macOS, and Windows, as well as auditable compilation scripts you can run yourself.

What FFmpeg is

FFmpeg is the industry-standard open source framework for audio and video processing. It handles decoding, encoding, transcoding, muxing, demuxing, streaming, filtering, and playback across virtually every media format in existence. Its core libraries — libavcodec, libavformat, libavfilter — form the foundation of countless broadcast, streaming, and media production systems worldwide.

Despite its ubiquity, the FFmpeg binary shipped by most Linux distributions and package managers is deliberately limited. Codec libraries with commercial or patent implications — most notably libfdk-aac, the reference AAC encoder and widely regarded as the best-quality AAC implementation available — are excluded from official packages. The result is an FFmpeg that covers basic use cases but cannot handle the codec requirements of professional audio workflows.

Why we use it

Audio quality is non-negotiable across every Fremen product. Whether encoding for distribution, resampling streams, or running real-time audio pipelines, we need AAC encoding that meets broadcast standards, MP3 output that is tuned correctly, and a resampler that does not introduce audible artifacts. Standard package-manager FFmpeg cannot deliver this.

So every Fremen product — across Linux servers, edge hardware, and containerized deployments — runs on this build. The same codec stack, the same build flags, the same verified binary. Having a single, reproducible FFmpeg means our audio output is consistent across the entire product line, and any update propagates everywhere at once.

The compilation scripts are fully auditable, version-pinnable, and designed for the environments we actually run: Debian/Ubuntu servers, Raspberry Pi edge nodes, macOS development machines, CI/CD pipelines, and minimal Docker images. We make them available publicly because maintaining a working FFmpeg build with the right codecs is a solved problem that should not cost every audio developer their afternoon.

What the pre-compiled binaries give you

Building FFmpeg from source takes 5 to 30 minutes, requires nasm, autoconf, pkg-config, a collection of -dev packages, and roughly 2 GB of working space. That is acceptable once. It is friction you do not want when provisioning infrastructure, bootstrapping CI runners, or keeping Docker images lean.

The pre-compiled binaries eliminate all of that. Built by CI for Linux x86-64, Linux ARM64, macOS ARM, macOS Intel, and Windows, each is a self-contained static executable with no external library dependencies. No installation, no ldconfig, no build environment. Drop it in, make it executable, and it works.

Every binary includes the complete audio codec stack our products depend on:

• libfdk-aac — High-quality AAC encoding, unavailable in standard packages
• libmp3lame — MP3 encoding
• libopus — Low-latency, high-quality codec for streaming and VoIP
• libsoxr — High-fidelity audio resampling

FFprobe is also included. For deployments that need more — shared linking, hardware acceleration, display output, manpages, or FFplay — the build scripts compile whichever variant fits your environment, with interactive version selection from FFmpeg 4 through the latest stable release.