Overview
- Smallest.ai closed a $13 million Series A led by Seligman Ventures that brings the company’s total funding above $21 million and will be used to scale commercial deployments.
- The company publicly launched Voice 4.0, built on an asynchronous Hydra architecture that runs a small, real‑time voice model in parallel with an offline LLM to manage turn‑taking and handoffs.
- Its named models include Pulse STT Pro for speech‑to‑text and Lightning V3.1 for text‑to‑speech, which the company says support 38 languages and enterprise features such as speaker diarisation, emotion detection and code‑switching.
- Smallest.ai trades some raw TTS blind‑vote naturalness for speed, claiming Lightning can generate 10 seconds of audio in about 100 milliseconds so agents respond with virtually no perceived lag.
- The startup lists early enterprise customers including RingCentral and Truecaller and plans to target contact centres, financial services, healthcare and BPOs as investors push into low‑latency audio infrastructure.