Assembly Powers Voice Apps Before, During and After Calls (Full Transcript)

[00:00:00] Speaker 1: Assembly is the voice infrastructure for building applications with spoken data. Let's take a single meeting. Let's say I want to schedule a meeting. I open the app called Blurred, which is built on top of our dictation API and speak to it. Hey, let's talk about our AWS instance for the meeting, especially the instance called c7gn.26large. And I send this message. So clean formatted text is sent and meeting is scheduled. Now, during the meeting, our real time speech to text API streams every word. So captions, note takers and agent assist never falls behind. With a universal family of models, the same setup keeps up across 32 languages. Follow speakers who switch languages mid sentence and labels who is speaking. Now talking about after the meeting, I have hours of transcripts on my phone. I send them to our pre-recorded endpoint and convert them to accurate transcripts. Speech understanding adds who spoke, how they felt and what mattered with summaries, chapters and translation. Guardrails redact the sensitive information before anything is shared. An LLN gateway sends the transcript to GPT, Claw, Gemini or open models to draft a follow up. And guess what? The next meeting, it's not me. It's our voice agent running it end to end. With our text to speech API, it carries a natural conversational voice. So every reply sounds like a person. So before, during, after voice infrastructure for every conversation. And so far, more than a million developers have built on top of a platform. On this YouTube channel, we share deep dives, tutorials and research behind our models. Try it for free on a playground at assembly.com slash playground and no credit card needed.