Users can now create longer music using the Lyria 3 Pro model, which is integrated into the latest version of Adobe’s Gemini. Users can now create tracks of 3 minutes, which is a huge improvement from the previously available 30 second tracks. With this new setup, the AI is able to complete songs with pre-defined structures, such as verses and choruses.
Key Takeaways:
- Gemini now produces AI music tracks that are 3 minutes long as opposed to 30 seconds.
- The Lyria 3 Pro model gives the ability to include additional elements such as verses, choruses, and bridges.
- This feature can be found in the Gemini app, Google Vids, and Vertex AI.
- Each song has a SynthID watermark to indicate that it is AI-generated.
With the Lyria 3 Pro model, users have greater flexibility to shape the sound of a song with the option to choose sections of the song that include an intro, verse or chorus. This feature enhances the overall music experience creating a more fluid sound rather than an arbitrary loop. The older versions of Lyria have inferior sound quality. This model is ideal for Youtube, video, or social media marketers requiring high-quality background music.
According to Google, background music creation has also improved. Google has also said that the model is an improvement of the understanding of the structure of music. Besides the Gemini app, the music creation tool will also be available in the upcoming Google Vids, Vertex AI, AI Studio, and Google Vids. This is beneficial for video developers and also for people creating videos.
To create the tool, Google partnered with music industry professionals. Yung Spielburg and DJ François K are music producers who have used Lyria in their work. Spielburg utilized it in a short film and François K used it to develop concepts for a new song.
Google is also working on copyright issues and safety. If you ask the AI to impersonate a specific artist, it will not do so. It will use the general feel or style of the artist to create a derivative work. This way, the AI is not accused of stealing an artist’s voice or style.
The company has implemented music originality detection. Users must abide by copyright and privacy protections. Each track created with Lyria 3 Pro is embedded with an unremovable SynthID watermark. This watermark serves to identify the file as AI-generated. Users may even query an audio file to Gemini to see if an AI watermark is present.
For Indian creators, this is a significant resource to produce materials for YouTube or Instagram reels. Creative audio for advertisements is also convenient for Indian small business owners. The tracks are of such quality that they sound good over both phone speakers and headphones.
FAQ
Q1: How long of music can Gemini generate now?
A1: Gemini can now generate music tracks of up to three minutes long! Previously this was limited to 30 seconds.
Q2: Can I request specific sections of the song, like a chorus?
A2: Absolutely. The Lyria 3 Pro model understands sections of songs. You can instruct Gemini to add a specific section like an intro, verses, chorus, or bridge.
Q3: Will the AI impersonate a specific artist’s voice if I ask?
A3: No. Google has placed restrictions on the AI to prevent it from imitating any specific artist’s voice or proprietary style. It will only take a general idea from the style of the track.
Q4: How do I know if a song was generated by Gemini?
A4: All songs generated by this tool will have a SynthID watermark. This is an AI-generated digital signature that certifies the song is AI-generated, and you can check this by uploading it to the Gemini app.
