a-r-t

Nice to see a Festool miter saw in his shop, Lynch knew what he was doing.

If we are speculating here, why not just go straight to an LLM serving all requests directly? No code needed.

Glad to hear it's on your radar. I'd imagine phone call transcription is a significant use case.

There are two options that I know of:

1. Merge both channels into one (this is what Whisper does with dual-channel recordings), then map transcription timestamps back to the original channels. This works only when speakers don't talk over each other, which is often not the case.

2. Transcribe each channel separately, then merge the transcripts. This preserves perfect channel identification but removes valuable conversational context (e.g., Speaker A asks a question, Speaker B answers incomprehensively) that helps model's accuracy.

So yes, there are two technically trivial solutions, but you either get somewhat inaccurate channel identification or degraded transcription quality. A better solution would be a model trained to accept an additional token indicating the channel ID, preserving it in the output while benefiting from the context of both channels.

Transcribing Twilio's dual-channel recordings using OpenAI's speech-to-text while preserving channel identification.

Hi Jeff, are there any plans to support dual-channel audio recordings (e.g., Twilio phone call audio) for speech-to-text models? Currently, we have to either process each channel separately and lose conversational context, or merge channels and lose speaker identification.

The MS announcement is limited to scams/phishing. Google mentions both scams and spam, but somehow I still get 15-20 spam emails a day that even the smallest LLM should be able to classify correctly.

You mentioned Django, but I couldn't find much on your website about how it works with your forms. Will I have to manually replicate all form fields on the client side? Will it work with Django form validation and show errors? I'd suggest creating a documentation page for each framework you mentioned to explain how it will work together.

The goal from the start was to raise $230M and then get acqui-hired for $116M by HP?

Off topic, but the font in the article is hard on the eyes.

From the article: "The researchers also applied their technique to three-millimeter cubes of brain tissue from a 9-month-old girl with epilepsy. The tissue maintained its pre-freezing structure and remained active in a laboratory culture for at least two weeks after being thawed."