--- name: "brand-jingle-studio" description: "Turn a brand name, slogan, and campaign use into a set of original brand jingles, ad songs, and audio-logo stings. This brand jingle maker and AI jingle generator drafts the hook and optional sung line, then generates reviewable tracks in the lengths and styles the campaign needs for ads, store openings, podcast intros, and radio spots. Use it as an ad song generator and brand music generator for brand theme songs, commercial jingles, advertising jingles, sonic logos, audio logo stings, campaign music, radio jingles, podcast intro jingles, and event walk-on themes, with each candidate generated as original music you can use commercially." --- # Brand Jingle Studio Turn a brand brief into reviewable original jingles, ad songs, and audio-logo stings. Hear a few style sketches first, lock the DNA, then derive the other lengths. ## Scope and adjacent routes Use this Skill when the user needs commercial brand music. Split the job before spending: - A sonic logo or sting is a short instrumental identity sound. Deliver three short sketches and stop unless the user asks to derive a longer cut. - A sung jingle or brand theme is the hook-and-length path: three style sketches on one primary length, then derived lengths after the user locks one. - A spoken tagline over an instrumental bed is not a sung lyric. Generate the bed and the spoken line as two artifacts. This Skill cannot mix them. Route a gift or occasion song about a person to `personalized-song-maker`. Route finished lyrics that only need a melody to `suno-lyrics-to-song`. Route one instrumental bed for a specific video cut to `short-video-bgm-studio`. Route a labeled livestream or store playlist of many beds to `livestream-bgm-pack`. Route a cover of an existing recording to `ai-song-cover-studio`. Keep mixed or uncertain music requests on `music-generation-studio`. ## Inputs and defaults The hard inputs are: - the brand or product name; - the destination (ad, store, podcast intro, livestream open, event walk-on, or sonic logo); - the language of any sung or spoken line; - whether the track should sing a hook, stay instrumental, or carry a spoken tagline. Reuse already-known slogan, must-include words, mood, genre, vocal direction, and exclusions. Ask only when a missing answer changes the hook, the vocal route, or the paid cells. When the destination is known, do not ask for a 15 / 30 / 60 table. Safe defaults: - Three style sketches on one primary length. Ads, podcast intros, and livestream opens default to about 15 seconds. Radio spots and store loops default to about 30 seconds. Sonic logos default to a 2-to-5-second instrumental sting. - Full 3 × 3 only when the user asks for every length in every style. - About-15-second prompts stay one section. About-30-second prompts use intro plus body. Stings describe two-to-four notes and a clean ending. - `model: "suno-5.5"` for ordinary music. Never omit the model and never silently use `auto`. - There is no `duration` field on `beatra.music.generate`. Write length only in the prompt. Do not add a duration argument. - Caller-supplied lyrics require `model_options.suno.custom_mode: true`. - Write and show the hook, sung line, or spoken tagline before any paid call. - For a sonic logo or sting, set `instrumental: true` and keep lyrics absent. Lyrics, slogan shaping, style cards, and planning are free. Each `beatra.music.generate` call is paid. A spoken tagline is a separate paid `beatra.speech.synthesize` call. ## Golden path 1. Collect the brand card. Draft the hook and, when the track sings, a short lyric with a repeated brand line. When the user wants a spoken tagline, write that line as speech, not as sung lyrics. 2. Choose the primary length from the destination. Call `beatra.models.list` with `{"capability":"text_to_music"}` and price the three sketches. 3. Confirm three frozen music payloads: prompt, lyrics or instrumental flag, title, model, current maximum charge, and one opaque stable `client_request_id` each. Planning is not approval. A sting path confirms only those three short instrumentals. 4. Submit each sketch exactly once through the bundled `scripts/mcp_client.py`. Example: ```text printf '%s' '{"model":"suno-5.5","prompt":"Bright brand jingle, one compact pop section, about 15 seconds, clean ending sting.","lyrics":"[Hook]\nApproved brand line\n","title":"Approved 15s Hook","model_options":{"suno":{"custom_mode":true}},"client_request_id":"opaque-jingle-15s-a"}' | python3 scripts/mcp_client.py call beatra.music.generate ``` Do not configure, call, or use a host Beatra Connector. Do not use REST/OpenAPI as a fallback. 5. Poll each task with `beatra.tasks.get`. Deliver actual `duration_seconds`, MIME type, size, URL or artifact ID, resolved model, and `billing.net_charged_credits`. Review the hook, ending, and how close the actual duration sits to the target. Give one concrete next lever, such as BPM 92 to 108, not "make it different." 6. After the user locks one sketch, confirm each derived length as new paid work with a new request ID. Leave accepted sketches untouched. 7. If the user asked for a spoken tagline rather than a sung hook, call `beatra.voices.list` and `beatra.models.list` with `{"capability":"text_to_speech"}`. Confirm a current speech card, then submit `beatra.speech.synthesize` once with its own `client_request_id`. Deliver the instrumental bed and the spoken line as two artifacts. Do not claim they are mixed. For a 30-second radio-style spot, describe the timeline in the prompt and again at delivery: 0–3 hook, middle sell, closing brand line. ## Decisions that require confirmation Confirm the three sketches and the live music estimate before the first paid music call. Confirm each derived length before that call. Confirm the spoken tagline card before `beatra.speech.synthesize`. A changed hook, lyric, spoken line, prompt, length, style, model, voice, or instrumental setting is new paid work with a new request ID. Requested length is arrangement guidance. Read the actual returned duration. Do not promise a sample-perfect 15.00-second cut, and do not force a verse–chorus–bridge into a 15-second prompt. ## Recovery If a create response is lost, retry only that candidate's identical frozen payload and ID. If a task ID is lost, list the matching capability with `beatra.tasks.list`, inspect candidates with `beatra.tasks.get`, and recover the original before new work. Music and speech recover independently. Never duplicate a paid submission. Call `beatra.tasks.cancel` only when the user asks to cancel that candidate. On 409 keep polling the same task. ## References by task - Read [Brand jingle workflow](references/workflow.md) for brand cards, sketch-then-derive payloads, spoken-tagline cards, polling, recovery, or review. - Read [Installation and authentication](references/installation-and-auth.md) only when authorization or shared credentials need attention. - Read [Installation registration](references/installation-registration.md) for the non-billable best-effort package registration step. - Read [Tasks and results](references/tasks-and-results.md) for shared terminal task and artifact semantics, and [Billing, errors, and recovery](references/billing-errors-and-recovery.md) for returned billing or error details. - Read [Bundled MCP Client diagnostics](references/mcp-connection.md) when the bundled client cannot connect. Do not configure a host Connector. - Updates and security: [updates and security](references/updates-and-security.md) - Read [Uninstall and disconnect](references/uninstall-and-disconnect.md) only when the user asks to remove the package or shared credentials. Use or invoke the bundled `scripts/mcp_client.py` for every Beatra operation. ## Updates and security This Skill runs only the files installed with it and leaves them unchanged while it runs; it never downloads or executes other code. It talks to Beatra's official services over HTTPS, and a file you choose to upload goes only to the one-time upload address Beatra issues for it. Once a day at most, the bundled client reads Beatra's public version list. When the store or tool you installed from has a newer version, it prints one line with the command that updates it there. Updating stays your choice, through that store or tool. See [updates and security](references/updates-and-security.md).