Typhoon Isan ASR Real-time
Real-time general Thai and Isan speech-to-text model with fast, accurate, and affordable deployment from laptops to enterprise.
Open-source Thai speech recognition model specialized for the Isan dialect, Typhoon Isan ASR Real-time delivers fast, accurate, and low-latency transcriptions of both Isan and general Thai speech while running efficiently on CPUs and compact GPUs. Fine-tuned from Typhoon ASR Real-Time on NVIDIA NeMo fastConformer-transducer-large architecture, it addresses the critical gap in ASR systems that struggle with regional Thai dialects, enabling true streaming without costly hardware or cloud dependencies. By preserving strong general Thai performance while adding robust Isan dialect recognition, Typhoon Isan ASR Real-Time democratizes accessible speech-to-text for Thailand's Northeastern region—from educational institutions and local businesses to government services and enterprise deployments serving diverse linguistic communities.