yousifa

This is super cool, will definitely try it out! Nice work

On the coreml side this is likely because the neural engine supports fp16 and offloading some/all layers to ANE significantly increases inference time and power usage when running models. You can inspect in the Xcode profiler to see what is running on each part of the device at what precision.

create whatsapp / signal group for each city and drop a user into it when they sign up

that already exists, its the microphone button

RIP legend. Not sure that I would be where I am today without Kevin being an inspiration and hero to me growing up.

If the startup has failed, it is highly unlikely that you would be able to get any $ from them. Even if you go after the founders personally, they likely don't have any cash and it will be a waste of time. If the company is still operating and has cash in the bank, 1000% you should go after them -- payroll is the first thing on the preference stack that gets paid when shutting down.

I will add, the fasting from food and water is the easy part. The more important fast is fasting from everything that is bad -- cursing, backbiting, lying, etc -- and from your own desires. All these things are things you shouldn't do anyway, but this is a time to refresh commitment to not doing those things.

As the Quran says: "Oh, you who believe! Fasting is prescribed to you as it was prescribed to those before you, that you may learn piety and righteousness" [Quran 2:183]

Eid mubarak!

thank you for your support!

Yes we are adding support for multi-language on the input and output sides. ASR will be switched to whisper later this week!

Thanks for the feedback. The next update of the app (tomorrow) will have many less outputs selected by default, thus less credits used. We definitely need an onboarding as well. Any suggestions on how we can introduce users to the experience?

We’ll make it happen!

might be something coming there soon ;)

TBD, but we may have something within a couple weeks. We're also working on a webapp.

Hey! We're thinking about adding a CarPlay app where you can just tap record on the screen (and tell siri to start a recording). Would that be helpful? Also, >10m recording length coming soon!

YES!!! This is what I want most too! We will be adding this soon.

Thats a great suggestion. We are changing the speech-to-text model we use later this week, so it should produce much better results for you. We will also be adding a feature where you can edit the transcript before submitting. Would that be helpful?

We don't have a sort order on that list right now, but it will be there within a couple days! We will also sort by your most used outputs.

You can already do that! Just deselect all the outputs on the main view (it will just give you a transcript in the history). Then in the history section, after you select the recording, you can click "Add Output" (or Add Response depending on the version of the app you have).

- nicely designed app, great work! - sliders on the right side are very difficult to use, instead of hover to select it would be better as click to select - have some sort of onboarding that lets me know I need to put in my api key

glad you're finding it useful! How can we make it even better for you?

we're currently using gpt-3.5 and gpt-4 for the LLMs. We'll be adding support for more models under the hood in the future depending on the prompt / use cases.

We’re experimenting and trying to do everything with the cheapest, smallest models we can, because ultimately we want to do it all on device

thanks for being a beta user! Where do you find yourself using Oasis most -- at your desk, on the go, etc? What subjects to you mainly use it for?

Thanks for trying it! Are there any outputs that you found to be particularly useful (or not useful)?

I highly recommend spending a couple hundred extra bucks and getting the macbook pro instead of the Air.

The Air has no fans and will throttle cpu extensively under medium - heavy usage. You will feel this and it is quite frustrating.

I also recommend you get at least 16GB RAM. I own a 2019 Mac Pro and my M1 16" MBP blows it out of this universe on every metric.

Docker runs fine, but you will need to add --Platform flag on some images to set the build platform as it will attempt to build for ARM if available. Other than that, it shouldn't require anything different.

Very cool use case! They all seem to be paintings / cartoons. Have you tried creating realistic human avatars?

It would also be cool to have an api where apps could use this to create default avatars for their users.

Blackhole lets you route audio.

https://existential.audio/blackhole/

Initial feedback:

1) Simplify the UI. Center the typing area and make it auto selected so I can type without having to click on it. 2) There is too much text. Just let me type without reading. 3) It seems like it requires full length questions to get good results. Focus on getting the same results with just a few words by inferring the question.

We’ve used Peter and he & his team are GREAT!

Background: They worked on an O-1 for one of our employees.

Point Of Presence (POP) is the location where two or more networks meet each other to connect.

Here is a good explanation: https://blog.stackpath.com/point-of-presence/