Podcast Voice Generator

OfflineTTS supplies generated speech tracks; it is not a multitrack podcast editor, remote recorder, hosting service, or distribution platform. Separate the script into intro, story sections, transitions, sponsor copy, and outro. Generate each section independently so timing, corrections, and music beds can be managed in an audio editor without rebuilding the whole episode.

Include host names, guest names, URLs, dates, abbreviations, and sponsor terminology in the first voice test. Never assume a brand name is pronounced correctly from spelling alone. Keep approved sponsor copy and the generation settings with each export, and obtain sponsor approval when synthetic delivery changes how a regulated or contractual statement sounds.

Review narration for cadence, factual accuracy, repeated words, clipped endings, and unnatural joins before mixing. A separate editor should handle music licenses, fades, loudness targets, chapter data, and the final feed file. WAV preserves more information for editing; MP3 is convenient for review but does not make an episode ready for every podcast platform.

Do not use a generated voice to impersonate a host, guest, or public figure. Confirm script and music rights, disclose synthetic narration where law, contract, or audience expectations require it, and choose an engine-language path consistent with confidentiality needs. Non-English Kokoro text may require a phonemization request even though final audio synthesis runs in the browser.