public static load(notnull repository as String, quantization as ModelQuantization = ModelQuantization.AUTO, notnull device as Device = Device.auto, revision as String = "main", offline as Boolean = false, verbose as Boolean = false, token as String = "", format as ModelFormat = ModelFormat.AUTO) as SpeechRecognizer
Downloads a verified GGUF pair or original Safetensors checkpoint. No repository name is hardcoded; architecture is validated from metadata.
| Parameter | Description |
|---|---|
repository | HF owner/name. @param quantization GGUF variant, or AUTO to retain original Safetensors precision/select an unambiguous GGUF Q8_0. |
device | CPU/NVIDIA/auto. @param revision Pinned branch/tag/commit. |
offline | Require an already-complete paired snapshot. |
verbose | Download/loading progress; default false. @param token Optional Hub token. |
format | AUTO, GGUF or SAFETENSORS; incompatible selectors are rejected. |
Owned reusable recognizer. @throws IOException for transfers/files; ValueException for mismatched checkpoints; NotSupportedException for another architecture or NPU. @example speech = SpeechRecognizer.load("unslothai/Qwen3-ASR-0.6B-GGUF", ModelQuantization.Q8_0)