Nari Qwen3-TTS and Qwen3-ASR: Top Accuracy, Lowest Latency and Cost (narilabs.com)

Nari Labs has announced that its Qwen3-TTS and Qwen3-ASR models have achieved top rankings in the latest Coval voice AI benchmarks, showcasing outstanding performance in both Text-to-Speech (TTS) and Speech-to-Text (STT) categories. Notably, the Qwen3-ASR Fast model ranks first in Time-to-Final-Segment (TTFS) with a latency of 44 ms and a Word Error Rate (WER) of 3.6%, while the Qwen3-TTS Fast model stands out with a TTFA of 63 ms and a WER of 3.8%. These metrics are critical for ensuring responsive voice agents that minimize user frustration and enhance overall experience. This achievement is significant for the AI/ML community as it highlights Nari Labs' commitment to advancing voice AI technologies with both high accuracy and low latency, placing them at the forefront of the industry. Additionally, the competitive pricing of their models—$0.12 per hour for STT and $10 per million characters for TTS—positions them as a cost-effective option compared to other models on the market. With the transition to a paid General Availability (GA) model coming soon, Nari Labs continues to encourage developers to explore their offerings through free trials, emphasizing innovation in voice AI capabilities.