The VoCore2 Ultimate has a sound card capable of recording and playing audio, so creating a small device for use with recent open-source AI tools could be a fun project.

• Sound Play The VoCore2 Ultimate uses the ES8388 sound card, which allows it to directly connect to an external 3.5mm standard speaker.
• Sound Record The ES8388 also supports audio recording functionality. On the VoCore2 Ultimate, I have already routed a microphone interface to a 3.5mm jack. For higher-quality recordings, you can instead use a USB microphone to achieve better results.
• AI Server This component cannot run directly on the VoCore due to high memory requirements—even for smaller models—but we can leverage open-source AI APIs like DeepSeek. I will test using the Qwen32B
• model, which works well on my laptop.

The idea works like this: First, record audio input and send it to a server to convert the sound into text. The text is then forwarded to an AI server for processing, where we wait for its response. Once the reply is received, the text is converted back into an audio file (e.g., WAV format), which is transmitted via Wi-Fi or Ethernet to the VoCore2 device. Finally, the audio is played through a connected speaker.

There’s no difficult part—just a few lines of code should be able to accomplish it. 😊 Maybe I can directly use FFmpeg on the VoCore2, so I won’t need to do most of the coding myself. Additionally, the AI itself will help with coding. Hopefully, I’ll finish this project by the end of this week!