Audio model · Google ·
Lyria 3.5 is an audio model by Google. In Endless it costs 16 credits per clip and is available on every plan.
Lyria 3.5 is Google DeepMind's music model: full songs with vocals, or instrumentals, from a text description and an optional image.
16 credits per clip
Lyria 3.5 is Google DeepMind's music generation model, launched in July 2026. Describe the track, with genre, mood, instruments, tempo, vocal style and structure, and it writes and performs a complete song, lyrics included, up to a few minutes long.
Write lyrics in the prompt to have them sung, or let the model write its own. Vocals work in English, German, Spanish, French, Hindi, Japanese, Korean and Portuguese. Turn on Instrumental for a track with no voice.
Attach an image and the music follows its mood and theme. There is no exact duration setting: ask for a length in the prompt, or mark sections with timestamps such as '[0:00-0:30] Intro'. Each track is an MP3 at 44.1 kHz stereo, carries Google's SynthID watermark, and costs the same whatever its length.
Start a chat in your workspace and ask for Lyria 3.5 by name.
Paste your script or describe the sound, its tone and its length.
Ask for another take or a different voice. Every clip is saved to your workspace assets.
by Google
Expressive speech in 30 voices, with the delivery directed in plain words
From 1 credit per clip
by MiniMax
Voiceovers with laughs, sighs and pauses written into the script
From 1 credit per clip
by ElevenLabs
Expressive narration and character voices in 90+ languages
Price depends on your input
Lyria 3.5 is on every plan, including Free. Create a workspace and get 100 credits to start, plus 50 credits every week.