Narration voice candidates

The same passage in each voice: part 1 of the bells episode. Pick one; the series switches to it with a --voice option.

Episode one, the bells. One component of the grid: the bells of the Peter and Paul belfry, which in the model set the frequency for the whole district. Nothing downstream runs without them. Here is what is documented, what the model claims, and what is still open. At the end we zoom out into the district they belong to.

• am_michael Kokoro-82M, free, local
• am_fenrir Kokoro-82M, free, local
• am_onyx Kokoro-82M, free, local
• am_adam Kokoro-82M, free, local
• am_echo Kokoro-82M, free, local
• bm_george Kokoro-82M, free, local
• bm_lewis Kokoro-82M, free, local
• bm_daniel Kokoro-82M, free, local
• bm_fable Kokoro-82M, free, local
• Gemini TTS, male prebuilt voices (Charon, Fenrir, Orus, Puck) HTTP 403 on the outstanding billing balance since 2026-11-05
• ElevenLabs, male voices paid; not recorded yet
• OpenAI gpt-4o-mini-tts (onyx, echo, fable, ash) paid; not recorded yet
• Higgsfield MCP voices (list_voices / create_voice) not recorded yet

Current renders use Kokoro af_heart (female). Recorded 9 candidates on 2026-11-08.