Hugging Face releases AnySpeechRecognizer and AnyDecisionModel Swift packages for Apple platforms

Hugging Face releases two open Swift packages, AnySpeechRecognizer and AnyDecisionModel, for running speech recognition and decision models on Apple devices.

AnySpeechRecognizer implements Apple's SpeechAnalyzer API and lets developers choose NVIDIA's Parakeet or Alibaba's Qwen3-ASR as the local Core ML recognition engine, and adds speaker diarization to track who speaks when. AnyDecisionModel produces typed yes-or-no, multiple-choice, and rating answers with probabilities from model outputs, supporting MLX, Core AI, llama.cpp, Jev, and Ollama backends.

Both packages require Swift 6.2 or later and macOS 26 or iOS 26 or later. The Parakeet and Sortformer runtimes and the MLX backend require Apple silicon. Model downloads occur only when an app requests them: about 600 MB per Parakeet, about 590 MB for the live encoder, 0.9 to 2.4 GB for Qwen3-ASR, and about 250 MB for Sortformer.