Whisper large-v3

OpenAI

Mhux ospitat — backend mock Piżijiet miftuħa

Sovranità u hosting

Mhux ospitat — backend mock

Mhux ospitat fuq LLM EU fuq dan id-deployment: it-talbiet jiġu mwieġba mill-backend mock integrat fir-reġjun dev-mock. L-ebda weights m'jaħdmu.

Sommarju

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.

Speċifikazzjonijiet

Fornitur
OpenAI (US)
Tieqa tal-kuntest
Mhux verifikat: il-fornitur ma dokumentax din id-dikjarazzjoni.
Parametri
1.54B
Modalitajiet
audio
Liċenza
apache-2.0 · licence
Piżijiet miftuħa
yes
Data tal-ħruġ
07 Nov 2023
Kompiti
translation, multilingual-support, summarization

Lingwi

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

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

Lingwi

Qawwi

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

Adekwat

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

Mhux verifikat: il-fornitur ma dokumentax din id-dikjarazzjoni. — a language is listed as strong only where a source claims it; the card links the evidence where one exists.

Meta tużaha

Saħħiet

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

Limiti

  • 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