What problem does it solve? Turning voice notes, audio clips, and video recordings into text usually means spinning up a separate Whisper process or writing custom HTTP calls. This Skill exposes the APX daemon's preloaded speech-to-text engine through a single CLI command, so transcription is fast, scriptable, and consistent. ## Core Features & Use Cases - File and folder transcription: Transcribe single files, multiple files, or entire folders of audio (webm, ogg, opus, m4a, aac, mp3, wav, flac) and video (mp4, mov, mkv, avi, and more), with ffmpeg extracting audio tracks automatically. - Provider routing: Choose between the local embedded Whisper (MLX on Apple Silicon, faster-whisper on NVIDIA, CPU otherwise), OpenAI's cloud Whisper, or any OpenAI-compatible custom STT endpoint via the --provider flag. - Scriptable output: Get clean transcripts on stdout for piping, or structured JSON with language, backend, and model fields via --json. - Use Case: You receive a folder of Spanish voice notes from a meeting. Run a bulk transcription with the language pinned to Spanish and pipe the JSON output into your notes pipeline. ## Quick Start Ask the agent to transcribe the file 'nota.oga' into text using the apx transcribe command.
