Four capability test results: song generation producing 175 seconds of audio, audio-to-score transcription of a 4-second source taking 36 seconds, sheet music rendering outputting a PDF and a playback audio file, and reference voice timbre completed in 47 seconds.