Tag: voxtral
All the articles with the tag "voxtral".
-
Chilean Spanish ASR Model Benchmark Report: Comprehensive Evaluation of Open-Source Speech Recognition Models
In-depth benchmark of 7 state-of-the-art ASR models on Chilean Spanish. Voxtral-Mini-4B achieves 2.63% WER, Whisper delivers 9.5x real-time speed, and Fun-ASR offers best cost-efficiency. Complete performance analysis, deployment recommendations, and roadmap.