Lyria 3.5 screens prompts: brand slogans and lyrics in an ad song
Google says every Lyria prompt is safety-screened and it cannot output copyrighted lyrics or named artist voices. How to write a brand jingle brief that passes.

If you ask Lyria 3.5 for a jingle, expect the prompt to be screened. Google's guide says all prompts go through safety filters and that Lyria cannot generate specific artist voices or copyrighted lyrics. So write your own original slogan lyrics, never paste a known song's words, never name a singer to imitate, and keep the musical brief intact if a request is rejected so the retry still describes the sound you want.
What the vendor says
The Gemini API music guide (read 2026-10-04) says all prompts are checked by safety filters and that prompts which trigger them are blocked, including prompts that request specific artist voices or copyrighted lyrics. It supports custom lyrics with structural tags such as [Verse], [Chorus] and [Bridge]. The Sume Music 1.0 page mentions section markers in the form [0:00-0:30] Intro: ... and says a policy rejection should be handled by revising the flagged content while keeping the musical brief.
| You want | Risk | Write instead |
|---|---|---|
| "Sing like [famous artist]" | Named artist voice is unsupported | A voice description: warm, close, mid-tempo female vocal |
| Lines from a hit song | Copyrighted lyrics are unsupported | Original lines about your product |
| A brand tagline as the chorus | Screened like any prompt; trademarks are your clearance | Keep the tagline short; clear it yourself |
| Aggressive competitor mentions | May be flagged | Leave competitors out of the prompt |
A brief that passes
Write the sound first and the words second. Describe emotion, tempo as a number, key, two to four instruments with texture, and one named moment in the arc. Then add the lyric lines you own. If the track should be a bed, close with "Instrumental, no vocals." and skip lyrics altogether, which removes the lyric clearance question from the pipeline.
- Use original lines, and clear the slogan separately from the audio.
- Keep the prompt under 5000 characters, the Sume limit.
- After a rejection, change only the flagged phrase.
What happens on Sume
Sume's Music Router takes the same kind of prompt and passes it to the engine, so the provider's screening still applies. Sume's docs say non-empty negative_prompt is unsupported (HTTP 400) and duration is rejected. Poll the job with Jobs and results; result.lyrics carries model-reported lyrics or a section map when the model provides one, which is metadata, not a measurement of the audio.
Closing note
Audio that passes screening is not automatically cleared for broadcast. A slogan can be someone's trademark, and a melody can resemble an existing song by accident. Listen before you ship, and keep the clearance notes with the asset.
Sources
Related posts
More in Use cases
- Make 3-minute hold music for a voice agent
ElevenLabs agents can play hold audio up to 180 s and 40 MB. Generate a short loop with Sume music generation and join it into 180 s with timeline audio.
- Meta ad video length by placement: 15 s to 240 min
Meta's placement chart sets different lengths and ratios per placement, from 15 s Messenger Stories to 240 min Feed. A render plan by placement.
- Meta's AI info label: signals or self-disclosure, and a trimmed clip
Meta applies its AI info label from industry-standard signals or self-disclosure. A trim or re-encode may drop file metadata, so disclose at upload too.
- Meta auto-detects AI-made ads: plan for the AI info label
Meta says its ad system detects ads made or edited with third-party AI tools through industry-standard signals and may show an AI info label. What to prepare.
Written by Sume