Made with Wavemaker on 2026-10-23 on staging: the chatterbox-dialog graph on one L40S in one job (38 billed GPU-seconds, 3.9 credits). Both reference voices were themselves made by Chatterbox (its English and Turbo built-in voices). Every line transcribes exactly; the voices sit at a median pitch of about 144 Hz and 222 Hz. The second speaker is about 7 dB quieter, following its reference, so the stems are normalised before mixing.

Wavemaker-authored example. The download holds the brief and the delivered media; model weights, graphs and tools are not included.
