VibeVoice tested by Argusic Runner

What the test found

The editable install succeeds, all 86 repository tests pass, the realtime TTS demo launches and serves HTTP 200/101, and real model inference produced a decoded non-silent audio file; real ASR inference could not load within the 8 GB memory limit.

Base `pip install -e .` succeeded; pip check reported no broken requirements. pytest passed 86 tests. The public realtime checkpoint generated a 1.73-second 24 kHz mono WAV (mean volume -22.4 dB), decoded by ffmpeg with exit 0. The launched demo served / and /config with HTTP 200 and /stream with HTTP 101. Real ASR preprocessing succeeded on two clips, but ASR model inference exceeded the 8 GB container limit. cp tests were not exercised; no vLLM/GPU stack was available.

Argusic installed VibeVoice in 2 minutes and launched it; verification reached Verified running for real. 4 of 5 recorded errors were worked through (see the timeline below).

Runsall runs of VibeVoice

This summary is drawn from the agent's recorded report for this run. Every figure traces to the log and recording above; nothing here is authored.