Optimize audio for transcription

Downmix to mono and downsample to 16 kHz to produce a clean, standard WAV for transcription. Smaller than uncompressed or video sources, and always lossless. Runs entirely on your device — your file is never uploaded.