compose_jingle
ChatGPTUse this first, when the user wants a jingle, station ident, podcast intro, DJ drop or short radio-style promo. From the brief it picks an intro, a music bed and an outro from AI Jingle Maker's catalogue, chooses a voice and sets the script. Free; nothing is charged. Each call creates a new, separate draft with its own draft_id; earlier drafts are left unchanged. Call it again only when the user wants a different jingle; to adjust the current one use swap_sound and set_script. Do not use it for music licensing questions, full songs, or audio longer than about 60 seconds.
get_jingle
ChatGPTFresh download links (24 hours) for a jingle the connected user rendered, by jingle_id. Read-only.
get_profile
ChatGPTReturn the connected AI Jingle Maker account: a stable opaque id, display name, email and current credit balance. Read-only; needs a connected account.
hear_sound
ChatGPTCard only: a 12-second watermarked clip of one catalogue sound in the draft.
list_voices
ChatGPTUp to 12 voices that can read the script, filtered by gender and language reach ('74 languages', '29 languages', '10 languages', 'English only'). Read-only. Use it when the user asks what voices exist or needs a voice for a language other than English; otherwise compose_jingle and set_script choose a voice by gender.
preview_jingle
ChatGPTRender a free preview of the whole draft: the real voice-over mixed with the three sounds, watermarked and played on the card. Use it when the user wants to hear the jingle before deciding. Not the deliverable: the clean MP3 and WAV come from render_jingle. Limited to a few per day per user.
render_jingle
ChatGPTProduce the final, clean MP3 and WAV of the draft. Spends credits from the user's connected AI Jingle Maker account: 1 credit for a script up to 125 characters, 2 up to 500 (same as the website); re-rendering an unchanged draft is free. Call only when the user clearly asks to render, buy, finalise or download. If the result says needs_confirmation, tell the user the cost and call again with confirm=true after they agree. If the user has no credits, explain that credits are managed on their AI Jingle Maker account page; never quote prices or link to a checkout. Whatever the outcome, finish the turn with show_jingle so the user sees the draft.
search_sounds
ChatGPTShortlist up to 5 catalogue sounds for one slot (intro, bed or outro) from a few words: mood, genre, instrument, pace, 'with a voice saying…'. Read-only; nothing is changed. With a draft_id the results are queued as that slot's alternates so swap_sound next/prev can cycle them. Use swap_sound with a brief when the user simply wants the slot replaced.
set_script
ChatGPTSet the voice-over text, the voice (by name, or by gender for a sensible default) and the reverb/chorus switches of an existing draft. Charges nothing. Nothing is lost: the previous script, voice and mix, and any jingle already rendered, are kept in the draft's history (and rendered jingles stay in the user's account), so an earlier version can always be restored by setting it again. Voices labelled 'English only' must not read a script in another language.
show_jingle
ChatGPTDisplay the draft as an interactive card: the three sounds with play and swap controls, the script, the voice, the watermarked preview player and, once rendered, the MP3 and WAV downloads. Call it exactly once, as the LAST step of a turn, after compose_jingle, swap_sound, set_script, preview_jingle or render_jingle have run; never more than once per turn and never before those tools. The card lets the user swap sounds, edit the script, change the voice and hear a preview by themselves, so after showing it keep the reply to one or two sentences.
swap_sound
ChatGPTReplace the intro, bed or outro of an existing draft. Give a direction (faster, slower, brighter, darker, calmer, more_energetic, shorter, longer, different), next/prev to cycle the queued alternates, or a few words to search (e.g. 'something with piano'). Free. Requires the draft_id from compose_jingle.
