Blog
All
Company updates
Research
Product
Case studies
Behind VAD 2.1: How we built our drive-thru benchmark
Synthetic noise can't replicate a real drive-thru. Here's how our recording team built the dataset that pushed VAD 2.1 the hardest.
Tyto 1.1: Sharper audio insight for every call
Tyto 1.1 delivers faster, more accurate audio-risk detection for the degraded call conditions that silently break speech-to-text and turn-taking in production voice agents.
Introducing VAD 2.1: Speech detection that holds up where production audio gets hard
VAD 2.1 delivers more reliable speech detection in the noisy, real-world conditions where production voice agents struggle most.
Introducing VAD Voice Focus 2.0: Voice activity detection for the primary speaker only
Built for production voice AI, VAD Voice Focus 2.0 triggers only on the primary speaker, ignoring background voices, TVs, and cross-talk that break classic turn-taking.
Introducing Quail Voice Focus 2.2: Primary speaker isolation in every situation
Built for production voice AI, Quail Voice Focus 2.2 delivers more robust speaker isolation across near-field, far-field, and multi-speaker scenarios.
Phonely x ai-coustics: Speaker isolation and audio insight for production Voice AI
See how Phonely improved transcription accuracy and qualification outcomes at scale with ai-coustics, treating audio quality as a core part of its voice AI infrastructure.
Introducing Tyto: Audio Insight into every call for Voice AI teams, at scale
Tyto helps voice AI teams identify audio issues before they become agent failures, with real-time and post-call insights into noise, packet loss, reverb, and more.
Engineering production grade voice agents with PolyAI and ai-coustics
Voice AI's hardest problem isn't sounding human - it's reliably hearing one. With ai-coustics, PolyAI cut false barge-ins by 40% and short-utterance failures by 30%.
Introducing the new VAD 2.0: Robust speech detection for real-time voice AI
VAD 2.0 is a standalone, noise-robust voice activity detector built for production voice AI - a drop-in Silero alternative that holds steady where energy-based VADs misfire.
