Mozilla is developing Whisperfile, a high-performance transcription AI based on OpenAI's Whisper



Mozilla, the developer of the web browser

Firefox , has announced that it is developing a transcription AI called Whisperfile , based on OpenAI's Whisper . Whisperfile is said to be able to translate audio in languages other than English into English.

Mozilla/whisperfile · Hugging Face
https://huggingface.co/Mozilla/whisperfile



Mozilla Developing Whisperfile For Local Audio-To-Text Translation - Phoronix

https://www.phoronix.com/news/Mozilla-Whisperfile

Mozilla's Whisperfile is based on the whisper.cpp software used in OpenAI's transcription AI 'Whisper'. Whisperfile can run on systems such as Linux, MacOS, Windows, FreeBSD, and OpenBSD. Whisperfile also works on both x86_64 and AArch64.

When you run Whisperfile, it automatically transcribes audio files in wav, mp3, ogg, and flac formats into text. It also supports color-coding to indicate the confidence level of speech recognition.



According to Mozilla, Whisperfile can also translate non-English audio into English.

You can also use the HTTP server by executing the following command.
[code]./whisper-tiny.en.llamafile[/code]



You can view the man page with the following command:
[code]./whisper-tiny.en.llamafile --help[/code]



Mozilla has published the Whisperfile source code and installation instructions in its official GitHub repository.

Whisperfile Documentation · GitHub

https://github.com/Mozilla-Ocho/llamafile/blob/6287b60/whisper.cpp/doc/index.md

in Software, Posted by log1r_ut