Whisper on RISC-V

Whisper is OpenAI’s open-source automatic speech recognition (ASR) model, widely used for on-device speech-to-text applications.

Category: AI / Speech Recognition

Status: Ongoing optimization and performance tuning for RISC-V AI PCs

Focus Areas:

• Porting Whisper inference to RISC-V CPU + NPU architectures
• ONNX optimization and quantization for RISC-V accelerators
• Improving real-time transcription performance on the DC-ROMA AI PC
• Preparing upstream patches for model runtimes and libraries

Why it matters: Whisper enables use cases such as offline transcription, live captioning, and intelligent media playback. Running Whisper locally on a RISC-V AI PC demonstrates that open instruction sets can support practical, everyday AI workloads.

Related resources:

https://github.com/DC-DeepComputing/Framework/blob/main/FML13V03/DC-ROMA%20RISC-V%20Mainboard%20II%20AI%20Model%20User%20Guide%20(Ubuntu%20AI%20Image).md