Whisper large-v3

OpenAI

No alojado — backend mock Pesos abiertos

Soberanía y alojamiento

No alojado — backend mock

No alojado en LLM EU en este despliegue: las solicitudes las responde el backend mock integrado en la región dev-mock. No se ejecuta ningún peso.

Resumen

Whisper large-v3 is OpenAI’s multilingual speech-recognition and speech-translation model with 1.54B parameters under the Apache-2.0 licence. It has a fixed 30-second receptive field and processes longer audio with a sliding window, and it uses 128 mel frequency bins rather than the 80 of earlier versions. The model card warns that it was evaluated on roughly ten languages and should be evaluated in-domain before deployment.

Especificaciones

Proveedor
OpenAI (US)
Ventana de contexto
No verificado: el proveedor no ha documentado esta afirmación.
Parámetros
1.54B
Modalidades
audio
Licencia
apache-2.0 · licence
Pesos abiertos
yes
Fecha de lanzamiento
07 nov 2023
Tareas
translation, multilingual-support, summarization

Idiomas

Fuerte: en, de, fr, es, it, pt, nl, pl

Adecuado: sv, da, fi, cs, ro, el, uk, ru, tr, ar, zh, ja, ko, hi

Idiomas

Fuerte

  • en
  • de
  • fr
  • es
  • it
  • pt
  • nl
  • pl

Adecuado

  • sv
  • da
  • fi
  • cs
  • ro
  • el
  • uk
  • ru
  • tr
  • ar
  • zh
  • ja
  • ko
  • hi

No verificado: el proveedor no ha documentado esta afirmación. — a language is listed as strong only where a source claims it; the card links the evidence where one exists.

Cuándo usarlo

Puntos fuertes

  • one model for transcription and translation to English
  • long-standing tooling support across every major inference stack
  • Apache-2.0 licence

Límites

  • fixed 30-second receptive field; longer audio needs chunking logic
  • the card only claims strong recognition for about ten languages
  • the card advises against using it for classification or speaker attributes