Try MiniMax FastH3 V2 from Fastvideo →
Sign InStart Building

Task History

Click to see output list

No tasks yet

Go to Models
Music GenerationActive

Lyria 3.5 Music Generation (Google)

bygoogle

Google’s Lyria 3.5 generates full-length songs from text prompts and optional images. It returns 44.1 kHz stereo MP3 audio plus lyrics text.

Text to SongImage to SongImage to Music
Model ID
lyria 3.5
Provider
google
Updated
1790326249
wiro playground—google/lyria 3.5
Reset to defaults
0 / 10
Maximum 10 image allowed
Delete All

Optional source images to inspire the music, up to 10 (jpg, jpeg, png, webp)

Required: Describe the full-length song (genre, instruments, mood, BPM, key, etc.). Set the length in the prompt (e.g. 'a 2-minute song') or use timestamps like [0:00 - 0:10] Intro. Use [Verse]/[Chorus]/[Bridge] tags for custom lyrics. Add 'Instrumental only, no vocals' for an instrumental track.

Sample outputs
Updated 1790326249

Overview

Lyria 3.5 is Google DeepMind’s flagship music generation model. It turns a detailed song description, and optional images, into a complete stereo track. It can also generate written lyrics for the vocal performance. The system uses latent diffusion over temporal audio latents to keep structure coherent across sections.

What you can build

  • Full song drafts for films, trailers, games, and podcasts
  • Photo-to-soundtrack concepts for mood boards, album art, and storyboards
  • Genre mashups with tight control over tempo, instrumentation, and dynamics
  • Vocal songs with generated lyrics, or instrumental-only backing tracks
  • Multiple versions of a theme, like different keys, BPM, or arrangement arcs

Inputs

  • A written description of the full song you want, as plain text. Include genre, mood, instruments, tempo (BPM), key, and arrangement notes.
  • Optional structure cues inside your text, like section tags (Intro, Verse, Chorus, Bridge) and clear transitions.
  • Optional timing guidance written into the description, such as timestamps for each section and a target total length.
  • Optional custom lyrics included directly in your text, clearly separated from instructions with a visible “Lyrics:” header.
  • Optional reference images to set mood and theme, up to 10 files in JPG, JPEG, PNG, or WebP format.
  • If you want lyrics in a specific language, write the request in that language.

Outputs

  • One audio file in MP3 format with 44.1 kHz stereo music, suitable for full song drafts up to about 3 minutes.
  • A lyrics text output when the track includes vocals. The text is typically grouped by sections, and it may include timing markers.

Limitations

  • Music generation is single-turn. You can’t edit a previous output by referencing it.
  • Duration control is prompt-driven. Exact length can vary from your target.
  • Outputs aren’t deterministic. The same request can produce different songs.
  • Vague or conflicting instructions can lead to generic structure or muddled genre intent.
  • Low-quality, blurry, or unrelated images can push the music in the wrong direction.
  • Safety filters may block requests for specific artist voices or copyrighted lyrics.

Safety & compliance

  • Follow Google’s Generative AI Prohibited Use Policy and applicable laws.
  • Don’t use the model to imitate a real artist’s voice or identity.
  • Don’t request copyrighted lyrics you don’t have rights to use.
  • Generated audio is watermarked with SynthID for content provenance.

Example prompts

Great starting points for lyria 3.5.

An epic cinematic orchestral piece about a journey home. Starts with a solo piano intro, builds through sweeping strings and choir, and climaxes with a massive wall of sound before fading back to solo piano. Around 2 minutes long.Music Generation
A dreamy indie pop song, 100 BPM, warm synths and soft female vocals, with the following lyrics: [Verse 1] Walking through the neon glow, city lights reflect below, every shadow tells a story, every corner, fading glory. [Chorus] We are the echoes in the night, burning brighter than the light, hold on tight, don't let me go, we are the echoes down below. [Verse 2] Footsteps lost on empty streets, rhythms sync to heartbeats, whispers carried by the breeze, dancing through the autumn leaves.Music Generation
[0:00 - 0:15] Intro: Soft lo-fi beat with muffled vinyl crackle and rain sounds. [0:15 - 0:45] Verse 1: A warm Fender Rhodes piano melody enters with gentle male vocals singing about a slow rainy morning. [0:45 - 1:15] Chorus: Full band with upbeat drums, bass and soaring synth leads. The lyrics are hopeful and uplifting. [1:15 - 1:45] Verse 2: Stripped back to piano and vocals, reflective and calm. [1:45 - 2:15] Final Chorus: Full band returns with layered harmonies. [2:15 - 2:30] Outro: Fade out with the piano melody alone.Music Generation
A high-energy synthwave track in A minor, 118 BPM, with punchy retro drums, pulsing arpeggiated bass, and a bright lead synth melody. Inspired by 80s night drives through a neon city. Instrumental only, no vocals.Music Generation

API quick start

Run lyria 3.5 with a single API call.

POST https://api.wiro.ai/v1/Run/google/lyria-3-5
{
  "prompt": "An epic cinematic orchestral piece about …",
  "inputImage": "https://your-cdn.com/input.png"
}
curl
curl -X POST "https://api.wiro.ai/v1/Run/google/lyria-3-5" \
  -H "Content-Type: application/json" \
  -H "x-api-key: YOUR_WIRO_API_KEY" \
  --data-binary @- <<'JSON'
{
  "prompt": "An epic cinematic orchestral piece about …",
  "inputImage": "https://your-cdn.com/input.png"
}
JSON
View full API docs