Running AI-Enhanced VLC Playback on the DC-ROMA RISC-V AI PC

As RISC-V advances from developer boards into full consumer-ready systems, one of the most important questions from users and developers is: Can RISC-V handle mainstream AI workloads?

To answer this, we built a proof-of-concept demo on the DC-ROMA RISC-V AI PC, powered by the ESWIN EIC7702 SoC with its 40-TOPS NPU. By integrating VLC media player, Whisper speech-to-text, and Qwen text-to-text processing, we demonstrated real-time AI-assisted media playback running natively on RISC-V.

This marks another milestone on our journey toward enabling complete AI workflows on open hardware.

Why VLC + Whisper + Qwen?

VLC is one of the world’s most widely used open-source media players. Adding real-time speech recognition and LLM interpretation turns it into an “intelligent media player” capable of:

• Generating subtitles for audio/video on the fly
• Summarizing or translating spoken content
• Enabling search inside long audio tracks
• Creating accessible playback for noisy environments or for the hearing-impaired

Running this end-to-end pipeline on RISC-V—without x86 or ARM acceleration—is a strong validation of the EIC7702 platform.

Demo Overview

The demo performs three AI tasks while playing a media file:

1. Real-time speech-to-text (STT) using Whisper

• Input audio frames are streamed from VLC
• Whisper (quantized ONNX model) performs inference
• Results are sent to the UI for live transcription

2. Text-to-text processing using Qwen
   Once STT output is ready, Qwen can:

• Summarize content
• Translate content
• Enhance or reformat text

3. Full inference pipeline running on RISC-V

• CPU: EIC7702’s 8-core RISC-V processors
• NPU: 40 TOPS accelerator for Whisper and Qwen
• OS: Ubuntu Desktop

This is one of the first public demos showcasing LLM and STT workloads running together on a consumer RISC-V laptop form factor. Please check the step-by-step demo through the link below:

https://github.com/DC-DeepComputing/Framework/blob/main/FML13V03/DC%20ROMA%20RISC-V%20AI%20PC%20VLC-STT%20Demo.md

Follow us on LinkedIn, X, and Facebook for updates, join our Discord community to connect and discuss, and visit our website to learn more about the RISC-V innovations.