I m Somil, I work on voice agents at Vaani in Bengaluru.
Something I read this week bugged me. One benchmark across 28 voice agents found the median one did no better than a plain words only pipeline. Meanwhile every new duplex model promises it hears emotion, sarcasm and hesitation that a transcript throws away.
Both can be true, but it made me wonder how much of what we call feels more human is just lower latency and better interruptions, and how much is the model understanding the caller better.
From my side, speech to speech models still seem less smart than the text LLMs they re built on. For anything involving tools, account lookups or a long conversation, I d still take the smarter model over the more natural one. So for people who ve tried both:
Own your audience and your income. Veamcast is a browser, an email client and a publishing platform in one Windows app. Send VEAMs: rich multimedia packages with video, audio, photos and links, straight to any inbox. Build channels, curate content from anywhere on the web, and earn from sponsors you pick, affiliates and paid subscriptions. Rate, tag and fact-check pages as you browse so the community, not an algorithm, decides what rises. No third-party trackers. Android app included.