dan

@kai_wen_eth @verdictprotocol @AssemblyAI hey @kai_wen_eth! for adding state persistence, e.g. if you want to continue from a past conversation, it's typically recommended to store past conversations on your end, and then inject the context into the prompt for the next session We're working on ways to make this easier!

@verdictprotocol @AssemblyAI what's the typical flow for adding state persistence to these one-hour agents? 👀

Voice agents are having a moment, and this Wednesday you can build one yourself, live, in about an hour. No coding experience required. Claude Code handles the heavy lifting. 🗓 Jun 10 · 10AM PT · Free

@alps_n_apps @alexcooldev @amosrazi Yes 🤝

@dan_aai @alexcooldev @amosrazi Solo bootstrapped founders are eligible?

I just received the bill for my app’s operational costs, now running at 👀: 💰 $19K/month 🫂 360K users 📈 5K–12K DAU Infra breakdown: - Cursor: $292 (Agent requests) - Supabase: $176 - Railway: $124 - OpenAI API: $392 - Speech-to-text: $428 - RevenueCat: $138 - Claude + Llama: $193 Infra total: $1,743/month (not including the 15% fees from the App Store and Google Play and 5% from Lemon Squeezy) I used to obsess over cutting infra costs. That was the wrong focus. I doubled down on marketing + distribution instead and that paid off. Market & build first. Optimize costs later.

@alps_n_apps @alexcooldev @amosrazi we have a startup program too :) so it could be $0 a month assemblyai.com/contact/startu…

@alexcooldev @amosrazi Assembly AI would save you a couple hundred $$$ each month here

trying something new - come chat! x.com/i/broadcasts/1…

Would you need to use the extracted email immediately (e.g. for booking a meeting)? If you can manage some delay or even wait until the end of the call, one common method we do see voice agent devs use, is transcribing the audio snippet containing the email with an async/batch model since they achieve better alphanumeric accuracy than realtime models Another thing you can try for a realtime experience is having the user spell out the email with the phonetic alphabet (A for Alpha, B for Bravo etc) as that can be more robust Overall every realtime model is not quite there yet for 95%+ accuracy across all providers, so it does require some workarounds, but it's possible to handle it gracefully and we humans need confirmation sometimes too

@dan_aai @OpenAI @AssemblyAI @elevenlabs @DeepgramAI Hey @dan_aai, thanks for the heads-up! Gonna test the multilingual streaming model today for extracting emails from German bookings. Any pro tips for tackling alphanumeric accuracy? Really appreciate your input!🙌

🇩🇪 Best STT for extracting emails from spoken German (e.g. restaurant voice bookings)? Tried 11labs, Deepgram, Vapi, Azure - 4/5 times garbage. Whisper? AssemblyAI? Custom fine-tune? Drop your real-world wins below, PLEASE! @OpenAI @assemblyai @elevenlabs @DeepgramAI

@gxespino Lol fair feedback, building the best performing speech to text models is always the main focus 😉

guys Assembly AI raised $50m in 2023 and they still don’t have team accounts today so, you can probably make do with whatever you have today

First month I've ever used up all my Windsurf credits Lots to ship

Launch day! Check out our new speech understanding, guardrails and LLM Gateway features today as well as a new docs redesign We just made it a whole lot easier to build Voice AI apps 😎

@uselayercode congrats team! 🔥

We're launching Layercode CLI to help you get started building voice AI agents in minutes 🚀 • Initialize a voice agent with a single terminal command • Test locally using our integrated tunnel • Deploy in minutes with our agent templates & sample backend • Easily connect to your existing backend logic & tools Just run ‘npx @layercode/cli init’ to get started.

plz like this tweet so we can convince our boss to buy this

this is 🔥🔥🔥 love to see agents working well with our API, the team has been grinding to improve docs readability for LLMs

@mattyp this is so fire, ffmpeg wasm is too quick 🔥🔥🔥 and glad to hear you like our API and that Replit agent can pick up the docs perfectly!

- ffmpeg wasm strips audio from video before processing -> this makes a 75m, 6gb video to about ~110mb - since ffmpeg wasm is local, the upload time is MUCH faster - I used AssemblyAI for transcription—it rips 🔥 - Replit App storage for storing audio / transcripts - Gemini Flash for transcript repair ffmpeg wasm is REALLY fun—here's a realtime look at how quickly it strips the audio for a ~75 minute, 6gb video to about 112mb

Transcribe a 75 minute video in under 2 minutes This app splits video from audio (locally), making upload size about 100x smaller Then, it uploads the audio to AssemblyAI for transcription Finally, transcripts are corrected with Gemini Flash Built with Agent 3, of course

@stevenvillarino @mattyp we support diarization :) assemblyai.com/docs/speech-to…

@mattyp Interesting. How does it handle diarization

if you were tasked with transforming a call centre with AI, how would you do it? here's the playbook i'd run: 3 stages 1 - start tracking your best performers with post call transcription and LLM as a judge analysis, you wanna start collecting the human data on what responses perform best and how questions should be answered, to power the next stages 2 - implement agent assist, use the data you collected from your best performers to power the agent assist responses. see how often agents are using the guidance, and call success rates to determine how well its performing 3 - add voice agents, once you know you're generating good responses and you have the right data, it's time to scale up with voice agents. you don't have to worry about improving agent responses as you already have the data from stage 1 & 2 and can focus just on making it sound natural most people try skipping straight to 3 and end up battling between improving responses and making it sound natural, hence why 95% of AI projects fail because you're jumping the gun no amount of prompting can make it sound human, you need to provide references of human responses

the ttfb (time to first byte) metric for measuring STT latency for voice agents is inaccurate, here's why you should be using ttct (time to complete transcript) instead: ttfb only measures silence -> start of speech -> first transcript chunk it completely ignores the more critical path which is ttct: end of speech -> silence -> final complete transcript (to be sent to the LLM) this can lead to some interesting conclusions: - streaming models that emit fast but inaccurate partials get rewarded (easy to game) - async models ran in a low latency fashion look like they perform terrible with ttfb due to no mid-speech partials (even tho they can work well in a voice agent) but the thing is: you're not using partials to generate the LLM response, you're using finals, so that is what you should measure models also have different configs for emitting partials vs finals, with partials you want to optimise for speed, finals you want to optimise for accuracy (without affecting latency too much) - so you will get different speeds at AssemblyAI we're laser focused on optimising for TTCT while maintaining industry leading accuracy, and we're shipping a new update today to our streaming model to return transcripts faster than any other provider via a new field called "Utterance" which emits finals on any 160ms silence so you can generate your LLM response as quick as possible - give your voice agents a noticeable speed up today by testing it out ;)

@icegiantx @pgrache @AssemblyAI @liveatc Sure thing! Dm me the URL :) @icegiantx

@dan_aai @pgrache @AssemblyAI @liveatc Would you be able to do this for me if I provide you the URL to transcribe? Could you put all of the text into a shared google docs document? Thank you @dan_aai @pgrache

Here I am trying to transcribe JFK Tower using the @AssemblyAI API (listenning to @liveatc). It's cannot really follow. I wonder if there are any models specifically fine-tuned for radio communications? After all, the phraseology and scope are relatively straightforward.

New improvements to @AssemblyAI's Universal-Streaming: - 21% accuracy improvement for repetitions in speech - 20ms faster response time - 66% better recognition of your custom keyterms - 3% improvement in accuracy - 4% better recognition for accented speech - 7% improvement at recognizing names, brands, and places

📣 Hey, look what’s new! Real-Time Call Transcription with FreeSWITCH and AssemblyAI: A Debugging Journey phparch.com/2025/09/real-t…

@icegiantx @pgrache @AssemblyAI @liveatc @icegiantx check out our getting started guide in our docs: assemblyai.com/docs/getting-s… if you need any support feel free to email our team: support@assemblyai.com or create a ticket via the popup in the bottom right of the docs :)

@pgrache @AssemblyAI @liveatc How do you get the @AssemblyAI API shown on the photo here? Thank you @pgrache