Theis Frøhlich

Random person sitting next to me has my extension installed 🥹

No feature update is too small... apparently.

Three years ago I quit my job inspired by people on this platform. What a rollercoaster it’s been. No regrets.

Tried a lot these this is definitely the best so far

👏👏👏

Going to keep this on for a while....

New ChatGPT Chrome Extension by OpenAI that replaces the default search engine in your browser. chromewebstore.google.com/detail/chatgpt…

Speech-to-speech API now available at $18 per hour 💸🤷‍♂️

Ok I'll just sit here and wait one week 🫠

I did a fun weekend project experimenting with AI video to create an ad for Talkio📽️😊 Base images: Midjourney Laughing faces: Luma Dream Machine Lip sync: Runway Voice and sound effects: ElevenLabs Text: ChatGPT Editing: Veed Idea: Human 😅

@testingcatalog Is this web or the app?

ChatGPT got a new icon for voice conversations. I found a good use case: learning German. If you ask ChatGPT to be your teacher, it can show words and how they’re written on the display. I think it’s super useful 🔥

@reach_vb I’m curious to see the pricing. Hume Evi does something similar and is priced at $6 per hour 🫣

I don't see how GPT4-o can be the go-to for all things audio: 1. It's an end-to-end model at the end of the day, but it doesn't have the same pitfalls as an AR Seq2Seq model, i.e., it'll hallucinate or make up transcriptions if it has low confidence in the audio or random speakers attribution for diarisation. 2. How does one verify if the transcription is correct at the end of the day? Do you employ a guidance step with a CTC model like WhisperX does? 3. Even though GPT4-o is 2x faster than GPT4, it would be significantly slower than Whisper over the API. Note: Whisper is just a 1.5B param model, as opposed to GPT4-o, which would be orders of magnitude larger. 4. On pricing, whisper on OAI API costs $0.006 / minute - there's no way you can match that on pricing with GPT4-o (in fact you'd incur a cost both on input and output tokens in this case - so your cost per minute would skyrocket) You can reduce the pricing even further for Whisper if you self host 5. Despite all of this, maybe you still want to use GPT4-o for transcription/ translation - you're still missing out on critical features like timestamps (word/ phrase), language probability, etc. Note: This is just my early assessment based on the blog post/ pricing page so it might be that some of these are invalidated, but I still don't see how it'd make sense to switch entire audio workflows to GPT4-o Lastly, I think it sets a pretty decent research direction for audio community, but for anything meaningfully usable in commercial settings we'd want it to be more feature rich.

@sairahul1 @MrNick_Buzz Me too, looks effective 😂

@MrNick_Buzz Bookmarking to implement in my saas 😜

Apps these days! Oh, you wanna cancel? We have a few questions for you... How about a discount? we are nice people Maybe not! answer one more question... One last time, just making sure you want to cancel... Oh, you left already? please come back 🥲

@rowancheung Technically this was filmed Wednesday last week. I was there so I know 😄

At a Q&A at Stanford University yesterday, Sam Altman commented on the future of AI and provided major insights. He called GPT-4 "mildly embarrassing at best", saying it will be the "worst model" we will ever use as each new version gets smarter.

Sam Altman just said GPT-4 is "the dumbest model any of you will ever have to use, by a lot" Plus, huge developments from Google Chrome, Scale AI, Anthropic Claude, and Ukraine's new AI spokesperson. Here's everything going on in AI right now:

@tdinh_me I have an entire copy of my nextjs app doing nothing but redirecting to www. Couldn’t get Namecheap redirects working either.

What's the easiest/fastest way to setup a redirect from Namecheap domain? 🤔 I swear to god the "Redirect Domain" feature in Namecheap dashboard never works.

@sobedominik Be careful about permissions! x.com/theisof/status…

I want to build a chrome extension 🧩 What are some common pitfalls I should expect and what can I do to avoid them? 🤔 I have seen horror stories of people needing to ship a hot-fix but it takes ages for the chrome store to take action etc.

@itsandrewgao @XDevelopers Do you know if the presentations are open to outside spectators?

Super excited to see the demos tomorrow afternoon @XDevelopers

Experimenting with Claude's Haiku model. It's surprisingly good considering its low price.

@marckohlbrugge I have a cron job running every hour for talkio.ai

If you run a SaaS with an annual plan, how do you handle monthly usage resets?

@idanmasas @PingParrot Great energy ❤️

I've spent approximately 128 hours building @PingParrot so far And it's literally just been 2 weeks. I'm gonna achieve my goals by the end of this year... - $10k in MRR - 10k followers on 𝕏 Watch me.