{"model_id": "nvidia/parakeet-tdt-0.6b-v3", "name": "parakeet-tdt-0.6b-v3", "owner": "nvidia", "source_url": "https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3", "task": "automatic-speech-recognition", "languages": ["en", "es", "fr", "de", "bg", "hr", "cs", "da", "nl", "et", "fi", "el", "hu", "it", "lv", "lt", "mt", "pl", "pt", "ro", "sk", "sl", "sv", "ru", "uk"], "tags": ["transformers", "nemo", "safetensors", "gguf", "parakeet_tdt", "feature-extraction", "automatic-speech-recognition", "speech", "audio", "Transducer", "Transformer", "TDT", "FastConformer", "Conformer", "pytorch", "NeMo", "hf-asr-leaderboard", "Transformers", "en", "es", "fr", "de", "bg", "hr", "cs", "da", "nl", "et", "fi", "el", "hu", "it", "lv", "lt", "mt", "pl", "pt", "ro", "sk", "sl", "sv", "ru", "uk", "dataset:nvidia/Granary", "dataset:nemo/asr-set-3.0", "arxiv:2509.14128", "arxiv:2505.13404", "arxiv:2305.05084", "arxiv:2304.06795", "arxiv:2410.01036", "arxiv:2406.00899", "arxiv:2205.12446", "arxiv:2012.03411", "arxiv:2007.10310", "arxiv:1510.08484", "license:cc-by-4.0", "model-index", "eval-results", "endpoints_compatible", "deploy:azure", "region:us"], "license_id": "cc-by-4.0", "license_status": "ouverte_avec_conditions", "license_label": "Creative Commons BY 4.0", "commercial_use": "possible_a_verifier", "gated": false, "private": false, "downloads": 601412, "likes": 1179, "updated_at": "2026-08-05T21:46:17.000Z", "description": "", "files": [{"name": ".eval_results/open_asr_leaderboard.yaml", "size": 2130, "format": "autre", "precision": "inconnue"}, {"name": ".gitattributes", "size": 1722, "format": "autre", "precision": "inconnue"}, {"name": "README.md", "size": 42888, "format": "autre", "precision": "inconnue"}, {"name": "config.json", "size": 1153, "format": "autre", "precision": "inconnue"}, {"name": "generation_config.json", "size": 289, "format": "autre", "precision": "inconnue"}, {"name": "model.safetensors", "size": 2508311120, "format": "safetensors", "precision": "inconnue"}, {"name": "parakeet-tdt-0.6b-v3.nemo", "size": 2509332480, "format": "autre", "precision": "inconnue"}, {"name": "parakeet-tdt-0.6b-v3.q8_0.gguf", "size": 713975456, "format": "gguf", "precision": "8-bit"}, {"name": "plots/asr.png", "size": 114075, "format": "autre", "precision": "inconnue"}, {"name": "processor_config.json", "size": 392, "format": "autre", "precision": "inconnue"}, {"name": "tokenizer.json", "size": 1159960, "format": "autre", "precision": "inconnue"}, {"name": "tokenizer_config.json", "size": 290, "format": "autre", "precision": "inconnue"}], "parameter_count": 627057286, "library_name": "transformers", "pipeline_tag": "automatic-speech-recognition", "datasets": ["nvidia/Granary", "nemo/asr-set-3.0"], "metrics": [{"name": "Test WER", "value": 11.31, "task": "Automatic Speech Recognition", "dataset": "AMI (Meetings test)"}, {"name": "Test WER", "value": 11.42, "task": "Automatic Speech Recognition", "dataset": "Earnings-22"}, {"name": "Test WER", "value": 9.59, "task": "Automatic Speech Recognition", "dataset": "GigaSpeech"}, {"name": "Test WER", "value": 1.93, "task": "Automatic Speech Recognition", "dataset": "LibriSpeech (clean)"}, {"name": "Test WER", "value": 3.59, "task": "Automatic Speech Recognition", "dataset": "LibriSpeech (clean)"}, {"name": "Test WER", "value": 3.97, "task": "automatic-speech-recognition", "dataset": "SPGI Speech"}, {"name": "Test WER", "value": 2.75, "task": "automatic-speech-recognition", "dataset": "tedlium-v3"}, {"name": "Test WER", "value": 6.14, "task": "Automatic Speech Recognition", "dataset": "Vox Populi"}, {"name": "Test WER (Bg)", "value": 12.64, "task": "automatic-speech-recognition", "dataset": "FLEURS"}, {"name": "Test WER (Cs)", "value": 11.01, "task": "automatic-speech-recognition", "dataset": "FLEURS"}, {"name": "Test WER (Da)", "value": 18.41, "task": "automatic-speech-recognition", "dataset": "FLEURS"}, {"name": "Test WER (De)", "value": 5.04, "task": "automatic-speech-recognition", "dataset": "FLEURS"}, {"name": "Test WER (El)", "value": 20.7, "task": "automatic-speech-recognition", "dataset": "FLEURS"}, {"name": "Test WER (En)", "value": 4.85, "task": "automatic-speech-recognition", "dataset": "FLEURS"}, {"name": "Test WER (Es)", "value": 3.45, "task": "automatic-speech-recognition", "dataset": "FLEURS"}, {"name": "Test WER (Et)", "value": 17.73, "task": "automatic-speech-recognition", "dataset": "FLEURS"}, {"name": "Test WER (Fi)", "value": 13.21, "task": "automatic-speech-recognition", "dataset": "FLEURS"}, {"name": "Test WER (Fr)", "value": 5.15, "task": "automatic-speech-recognition", "dataset": "FLEURS"}, {"name": "Test WER (Hr)", "value": 12.46, "task": "automatic-speech-recognition", "dataset": "FLEURS"}, {"name": "Test WER (Hu)", "value": 15.72, "task": "automatic-speech-recognition", "dataset": "FLEURS"}], "github_url": "", "github_metadata": {}, "paper_url": "https://arxiv.org/abs/2509.14128", "paper_doi": "", "paper_metadata": {}, "card_complete": true, "safetensors": true, "provenance": [{"source": "Hugging Face Hub API", "url": "https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3", "retrieved_at": "2026-10-09T01:52:35.626183+00:00"}], "retrieved_at": "2026-10-09T01:52:35.626183+00:00", "warnings": ["La licence est déclarée par le producteur ; vérifiez toujours son texte intégral et les conditions associées."], "rank_score": 0.0, "rank_reasons": [], "compatibility": {"state": "compatible", "label": "Compatible selon l’estimation", "summary": "La mémoire indiquée dépasse l’estimation prudente.", "factors": ["Mémoire estimée : 3.0 Go, marge d’exécution de 30% incluse.", "Espace disque minimal indicatif : 2.3 Go.", "Une exécution CPU peut être possible mais sa vitesse ne peut pas être déduite de ces métadonnées."], "estimate": {"known": true, "precision": "float32", "source": "taille réelle déclarée des fichiers de poids", "base_bytes": 2508311120, "estimated_bytes": 3260804456, "overhead": 0.3}}, "quantization": {"available": true, "variants": [{"label": "Q8_0", "filename": "parakeet-tdt-0.6b-v3.q8_0.gguf", "size": 713975456, "size_human": "681 Mo", "memory_bytes": 928168092, "memory_human": "885 Mo", "note": "Pratiquement sans perte perceptible.", "rank": 30, "memory_gb": 0.9, "state": "tient", "verdict": "Tient dans la mémoire déclarée"}], "recommended": {"label": "Q8_0", "filename": "parakeet-tdt-0.6b-v3.q8_0.gguf", "size": 713975456, "size_human": "681 Mo", "memory_bytes": 928168092, "memory_human": "885 Mo", "note": "Pratiquement sans perte perceptible.", "rank": 30, "memory_gb": 0.9, "state": "tient", "verdict": "Tient dans la mémoire déclarée"}, "lightest": {"label": "Q8_0", "filename": "parakeet-tdt-0.6b-v3.q8_0.gguf", "size": 713975456, "size_human": "681 Mo", "memory_bytes": 928168092, "memory_human": "885 Mo", "note": "Pratiquement sans perte perceptible.", "rank": 30, "memory_gb": 0.9, "state": "tient", "verdict": "Tient dans la mémoire déclarée"}, "memory_gb": 8.0, "disk_gb": 20.0, "fitting_count": 1, "overhead": 0.3, "note": "Tailles issues des fichiers déclarés par le producteur, majorées de 30% pour l’exécution. Une variante plus compressée consomme moins de mémoire et dégrade les réponses : cette dégradation n’est pas mesurée ici."}, "editorial_quality_score": null, "editorial_quality_blockers": []}