AI Music Generator / Lyria 2
Lyria 2 Online - AI Music Generator
Use Lyria 2 inside QuestStudio's AI Music Generator. Generate results, iterate with prompts, and export your outputs fast.
Best for
- Short instrumental beds
- Musical concept exploration
- Background textures
- Video soundtrack ideas
Key features
- Text-to-instrumental music
- Mood and instrumentation direction
- Short clips for review and editing
Limitations
- Does not generate lyrics or vocals
- Requested tempo and loop boundaries still need review
Sample prompts
Warm acoustic instrumental, restrained percussion, space for narrationGentle piano and soft strings, reflective mood, no lead vocal
Related models
Turn a Lyria 2 instrumental into a usable background cue
Lyria 2 generates instrumental audio, not a sung lyric. Its documented output is a 30-second clip. Start with the scene and the role of the music: a quiet bed beneath a speaker needs a different arrangement from a product reveal. The useful result is the passage that fits the finished edit, not simply the first audio file that sounds polished.
| Creative job | Start with | Decision rule |
|---|---|---|
| A quiet bed beneath narration | Open Lyria 2 | Describe restrained instrumentation, energy and room for speech. Review the returned clip with the actual narration; instrumental music can still compete with a voice through melody, rhythm or density. |
| An approved line must be sung | Use Music 2.6 | Use a vocal-capable model and enter the approved words in its lyrics field. Adding those words to a Lyria prompt does not turn an instrumental model into a singer. |
| A short cue must repeat cleanly | Check the loop boundary | Choose a complete musical phrase in the downloaded audio, inspect its boundary and compare repeated playback. A requested tempo does not certify a sample-accurate loop or a finished ending. |
A controlled workflow that wastes fewer generations
- Write the edit requirement before the prompt. In a fictional twelve-second tabletop demonstration, a spoken instruction runs through the middle and a product name closes the piece. Reserve room for those words; do not ask the music to fill every moment with a new lead phrase.
- Try an original brief such as: Warm plucked instrumental, restrained percussion, gentle forward motion, sparse melody, space beneath a calm spoken instruction. This is a proposed brief, not a tested model result. Select Google Lyria 2 explicitly and review account access before generating.
- Listen to the complete returned clip, then choose a passage that supports the scene. Compare versions at similar loudness. Mark the first useful beat, any distracting fill and a possible ending. Keep the source clip unchanged while making an edit copy.
- Fit the selected passage in an audio or video editor. For the twelve-second example, trim or arrange the music around the actual narration and final message. Check the beginning, any repeated join and the ending after export. A 30-second source does not need to become a 30-second ad.
Questions before you generate
Can Lyria 2 sing my advertising line?
No. This version generates instrumental clips. For a sung line, choose a vocal-capable model such as the current MiniMax Music 2.6 option and review every generated word.
Can I select a twelve-second output or a guaranteed loop?
The Lyria 2 provider input does not expose a duration or loop switch. Its documented clip length is 30 seconds. Select and edit the useful passage afterward, then check the actual boundary and final duration.
Can I upload a melody reference in this Lyria interface?
The current Lyria selection in Music Lab uses a text prompt and does not expose a reference-audio upload. Do not assume that a control described for another model or Lyria version is available here.
How should I compare two instrumental candidates?
Use the same picture and narration, similar playback loudness and a written acceptance list. Check speech clarity, distracting changes, timing and the ending. Record the selected passage; no universal listening-quality winner is claimed here.