Aaro on-device speech models (browser delivery copy)

Files the Aaro website downloads into the browser for on-device Hindi and English voice. Speech is processed on the student's device; nothing here runs on a server.

  • 53 files, 297.6 MB in total.
  • Each file is named by its SHA-256: <sha256> is raw bytes, <sha256>.bin is a gzip payload whose own SHA-256 is pinned separately.
  • The website checks the size and SHA-256 of every file before using it, and loads them through a URL pinned to one commit of this repository.
  • hi-IN: bundle pair-hi-IN-3b2adc22ab7d9cc054ed4f42
  • en-IN: bundle pair-en-IN-92cf677c86d965e57b8d432d

Licences and attribution

The models come from different authors under different licences (MIT, Apache-2.0, CC-BY-4.0, public domain) and include eSpeak NG data (GPL-3.0). Authors, upstream revisions, changes made and source offers are listed in NOTICE.txt. GPL-3.0 corresponding source (eSpeak NG and the workers that run it) is in source/. Licence texts: Apache-2.0.txt, CC-BY-4.0.txt, Moonshine-LICENSE.txt, Moonshine-runtime-THIRD_PARTY_NOTICES.txt, Whisper-MIT.txt, espeak-GPL-3.0.txt.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support