{"id":"nvidia-magpie-multilingual-357m-feb-2026","nom":"Magpie-Multilingual 357M (Feb 2026)","editeur":"NVIDIA","pays":"États-Unis","poids_ouverts":true,"poids_selon":null,"licence":"NVIDIA Open Model License","licence_selon":null,"liens":{"page":"https://huggingface.co/nvidia/magpie_tts_multilingual_357m","essayer":"https://huggingface.co/spaces/nvidia/magpie_tts_multilingual_demo"},"fiche_auto":true,"page":"https://quelleia.com/voix/modeles/nvidia-magpie-multilingual-357m-feb-2026/","releve":"2026-10-19","en_bref":"9e sur 16 en prononciation du français ; 54e sur 96 en synthèse vocale en anglais (Artificial Analysis), ex æquo avec 12 autres modèles dont les marges d'erreur se recouvrent.","presentation":"Modèle de synthèse vocale multilingue à poids ouverts de NVIDIA qui fait parler 5 voix dans 9 langues dans sa version de février 2026 (v2602). Il prédit des jetons de codec audio avec un transformeur encodeur-décodeur, et le clonage de voix zéro-shot a été retiré de cette version.","sortie":"2026-03-03","taille":"357 millions de paramètres","langues":{"annoncees":{"annoncees":"9 langues, dont le français","nombre":9,"francais":true,"liste":["en","es","de","fr","vi","it","zh","hi","ja"]},"prises_en_charge_selon_open_tts":9,"evaluees":["fr","en"]},"preuves":[{"appuie":["langues","presentation"],"lu":"https://huggingface.co/nvidia/magpie_tts_multilingual_357m/raw/v2602/README.md (« Each speakers can speak nine different languages (En, Es, De, Fr, Vi, It, Zh, Hi, Ja). »), 17/10/2026"},{"appuie":["presentation"],"lu":"https://huggingface.co/nvidia/magpie_tts_multilingual_357m/raw/v2602/README.md (« The model predicts discrete audio codec tokens autoregressively using a transformer encoder-decoder architecture. »), 17/10/2026"},{"appuie":["presentation"],"lu":"https://huggingface.co/nvidia/magpie_tts_multilingual_357m/raw/v2602/README.md (« We have removed zero-shot capabilities of this model for this release. »), 17/10/2026"},{"appuie":["sortie"],"lu":"https://huggingface.co/nvidia/magpie_tts_multilingual_357m/raw/v2602/README.md (« March 3, 2026: MagpieTTS v2602 was released with support for 2 new languages (Hindi and Japanese). »), 17/10/2026"},{"appuie":["taille"],"lu":"https://huggingface.co/nvidia/magpie_tts_multilingual_357m/raw/v2602/README.md (« Number of model parameters 3.57*10^8 »), 17/10/2026"}],"classements":[{"id":"synthese/prononciation","nom":"Prononciation du français","rang":9,"sur":16,"valeur":7.03,"unite":"%","chiffres":{"rtfx":8.96,"taille_milliards":0.31,"langues":9,"clonage":false},"voisins":["openmoss-team-moss-tts","meta-seamless-m4t-v2-large"],"source":{"nom":"Open TTS Leaderboard","mention":"Source : Open TTS Leaderboard, Hugging Face","url":"https://huggingface.co/spaces/hf-audio/open_tts_leaderboard","releve":"2026-10-19"}},{"id":"synthese/artificial-analysis","nom":"Synthèse vocale en anglais, Artificial Analysis","rang":54,"sur":96,"valeur":1063.98,"unite":"Elo","bas":1051.98,"haut":1075.98,"votes":2231,"chiffres":{"date_sortie":"2026-02-27"},"voisins":["kokoro-82m-v1-0","amazon-polly-generative"],"source":{"nom":"Artificial Analysis","mention":"Source : Artificial Analysis","url":"https://artificialanalysis.ai/text-to-speech/model-families/nvidia","releve":"2026-10-19"}}],"noms_dans_les_sources":{"hf_audio":"nvidia/magpie_tts_multilingual_357m","artificial_analysis":"Magpie-Multilingual 357M (Feb 2026)"}}