from Hacker News

Federated finetuning of Whisper on Raspberry Pi 5

by danieljanes on 11/16/23, 7:30 PM with 20 comments

• by filterfiber on 11/16/23, 7:50 PM

  I don't think the article mentions it, how well does the rpi 4 and 5 do for inference with whisper especially v3?

• by ulnarkressty on 11/16/23, 8:00 PM

  How would this actually work in practice? Do I ask the user to utter specific words then train on that? How is it different from the traditional speech recognition that I need to 'train' to work better on my voice?

  The Holy Grail would be to train the model while using it, without any friction. I don't think these methods support that though.

• by saqadri on 11/16/23, 8:41 PM

  This is cool. This might be a silly question, but what are the scenarios where it's useful for fine-tuning on the edge with small devices? I get inference on the edge, and curious about metrics on that for Whisper, but isn't it better to fine-tune on beefier infrastructure and then deploy it for inference on the edge?

• by Havoc on 11/16/23, 7:58 PM

  I’m guessing this will also help with thick accents?