argmax-oss-swift

Why we included this project

Speech recognition, speaker diarization, and text-to-speech usually mean sending audio to a cloud API, but Argmax's open-source SDK keeps all three on-device in a single Swift package for macOS and iOS. Whisper handles speech-to-text, Pyannote handles speaker diarization, and Qwen handles text-to-speech, so a full transcription pipeline with speaker labels can run entirely on Apple Silicon without shipping audio anywhere. That combination matters for privacy, offline use, and avoiding per-minute API charges, and it saves teams the work of stitching together several network calls. It is also a good way to gauge how far local speech AI has come on Apple hardware before committing to a hosted service.