{ "family": "owsm", "schema_version": 1, "display_name": "OWSM", "description": "ESPnet Open Whisper-style multilingual speech recognition and speech translation.", "category": "asr", "status": "supported", "tasks": [ "asr" ], "modes": [ "offline", "streaming" ], "languages": [ "abk", "afr", "amh", "ara", "asm", "ast", "aze", "bak", "bas", "bel", "ben", "bos", "bre", "bul", "cat", "ceb", "ces", "chv", "ckb", "cmn", "cnh", "cym", "dan", "deu", "dgd", "div", "ell", "eng", "epo", "est", "eus", "fas", "fil", "fin", "fra", "frr", "ful", "gle", "glg", "grn", "guj", "hat", "hau", "heb", "hin", "hrv", "hsb", "hun", "hye", "ibo", "ina", "ind", "isl", "ita", "jav", "jpn", "kab", "kam", "kan", "kat", "kaz", "kea", "khm", "kin", "kir", "kmr", "kor", "lao", "lav", "lga", "lin", "lit", "ltz", "lug", "luo", "mal", "mar", "mas", "mdf", "mhr", "mkd", "mlt", "mon", "mri", "mrj", "mya", "myv", "nan", "nep", "nld", "nno", "nob", "npi", "nso", "nya", "oci", "ori", "orm", "ory", "pan", "pol", "por", "pus", "quy", "roh", "ron", "rus", "sah", "sat", "sin", "skr", "slk", "slv", "sna", "snd", "som", "sot", "spa", "srd", "srp", "sun", "swa", "swe", "swh", "tam", "tat", "tel", "tgk", "tgl", "tha", "tig", "tir", "tok", "tpi", "tsn", "tuk", "tur", "twi", "uig", "ukr", "umb", "urd", "uzb", "vie", "vot", "wol", "xho", "yor", "yue", "zho", "zul" ], "capabilities": { "asr": [ "segments" ] }, "dependencies": [], "options": { "request": [ { "name": "language", "type": "string", "description": "Input language as an OWSM ISO 639-3 code, such as eng, jpn, or zho; auto detects it from the first window.", "required": false, "default": "eng" }, { "name": "target_language", "type": "string", "description": "Optional OWSM translation target code. Omit for transcription.", "required": false }, { "name": "return_timestamps", "type": "bool", "description": "Return OWSM utterance-level timestamp segments.", "required": false, "default": false }, { "name": "condition_on_previous_text", "type": "bool", "description": "Condition each long-form chunk on the previous chunk transcript.", "required": false, "default": false }, { "name": "num_beams", "type": "int", "description": "Decoder beam width; one selects greedy decoding.", "required": false, "min": 1, "max": 50002, "default": 1 }, { "name": "max_tokens", "type": "int", "description": "Maximum generated tokens per 30-second window; zero uses the model limit.", "required": false, "min": 0, "max": 374, "default": 0 }, { "name": "audio_chunk_mode", "type": "enum", "description": "Long-form audio chunking mode.", "values": [ "auto", "fixed", "none" ], "required": false, "default": "auto" }, { "name": "audio_chunk_duration_sec", "type": "float", "description": "Maximum duration of each audio window. Must be positive.", "required": false, "max": 30, "default": 30 } ], "session": [ { "name": "weight_type", "type": "enum", "description": "Matmul weight storage override.", "values": [ "native", "f32", "f16", "bf16", "q8_0", "q4_0", "q4_1", "q5_0", "q5_1", "q2_k", "q3_k", "q4_k", "q5_k", "q6_k" ], "required": false, "default": "native" } ], "load": [] }, "runtime": { "tags": [ "gguf" ] }, "ui": { "tags": [ "ASR", "GGUF" ], "docs": [ "docs/models/owsm.md" ], "recommended_package": "owsm_v4_base_102m_f32" }, "package_defaults": { "download": { "kind": "huggingface_snapshot", "repo": "audio-cpp/audio.cpp-gguf", "revision": "main", "gated": false } }, "packages": [ { "id": "owsm_v4_base_102m_f32", "display_name": "OWSM v4 Base 102M F32", "default": true, "format": "gguf", "precision": "f32", "target_directory": "OWSM-GGUF", "files": [ "OWSM-GGUF/owsm-v4-base-102m-f32.gguf" ], "strip_prefix": "OWSM-GGUF" }, { "id": "owsm_v4_base_102m_q8_0", "display_name": "OWSM v4 Base 102M Q8_0", "default": false, "format": "gguf", "precision": "q8_0", "target_directory": "OWSM-GGUF", "files": [ "OWSM-GGUF/owsm-v4-base-102m-q8_0.gguf" ], "strip_prefix": "OWSM-GGUF" }, { "id": "owsm_v4_small_370m_f32", "display_name": "OWSM v4 Small 370M F32", "default": false, "format": "gguf", "precision": "f32", "target_directory": "OWSM-GGUF", "files": [ "OWSM-GGUF/owsm-v4-small-370m-f32.gguf" ], "strip_prefix": "OWSM-GGUF" }, { "id": "owsm_v4_small_370m_q8_0", "display_name": "OWSM v4 Small 370M Q8_0", "default": false, "format": "gguf", "precision": "q8_0", "target_directory": "OWSM-GGUF", "files": [ "OWSM-GGUF/owsm-v4-small-370m-q8_0.gguf" ], "strip_prefix": "OWSM-GGUF" }, { "id": "owsm_v4_small_370m_q4_k", "display_name": "OWSM v4 Small 370M Q4_K", "default": false, "format": "gguf", "precision": "q4_k", "target_directory": "OWSM-GGUF", "files": [ "OWSM-GGUF/owsm-v4-small-370m-q4_k.gguf" ], "strip_prefix": "OWSM-GGUF" }, { "id": "owsm_v4_medium_1b_f32", "display_name": "OWSM v4 Medium 1B F32", "default": false, "format": "gguf", "precision": "f32", "target_directory": "OWSM-GGUF", "files": [ "OWSM-GGUF/owsm-v4-medium-1b-f32.gguf" ], "strip_prefix": "OWSM-GGUF" }, { "id": "owsm_v4_medium_1b_q8_0", "display_name": "OWSM v4 Medium 1B Q8_0", "default": false, "format": "gguf", "precision": "q8_0", "target_directory": "OWSM-GGUF", "files": [ "OWSM-GGUF/owsm-v4-medium-1b-q8_0.gguf" ], "strip_prefix": "OWSM-GGUF" }, { "id": "owsm_v4_medium_1b_q4_k", "display_name": "OWSM v4 Medium 1B Q4_K", "default": false, "format": "gguf", "precision": "q4_k", "target_directory": "OWSM-GGUF", "files": [ "OWSM-GGUF/owsm-v4-medium-1b-q4_k.gguf" ], "strip_prefix": "OWSM-GGUF" } ], "sources": [ { "format": "safetensors", "roots": { "model": "." }, "files": { "config": "model:config.json", "tokenizer": "model:bpe.model" }, "tensors": { "weights": "model:model.safetensors" } }, { "format": "gguf", "roots": { "model": ".", "weights": "$gguf" }, "files": { "config": "model:config.json", "tokenizer": "model:bpe.model" }, "tensors": { "weights": "weights:" } } ] }