VoxCPM TTS — KVM4 (CPU-only)
CPU inference: ~tens of seconds per phrase. First request downloads the model (~couple min).
Hello from VoxCPM running on the KVM4 server.
Synthesize