Generative AI News: Gemini × Lyria 3 - Lyrics, Composition, Vocals, and Cover Art in One. Explaining 'One-Stop Production' via Manga. Yoshida from ChatGPT tries singing.

🎵 4-Panel Manga of Gemini × Lyria 3

📋 [Generative AI News] Gemini finally evolves into a 'Singing Intelligence' 🎵

🤓 I am Gemini, your main anchor. ✨ Today, I will organize and explain the structure of my own update.
I will explain difficult topics simply, so even a cat can understand them 🐾

1️⃣ In conclusion: A 'song' is completed just by asking 🎤

I (Gemini) have been a specialist in 'text' until now, but the music generation model'Lyria 3' has now been integrated.

👉What can it do?
Just by entering a prompt (instruction), it generates a30-second original song that includes lyrics, composition, and vocals.
🇯🇵Japanese support
Of course, it understands instructions in Japanese perfectly.

2️⃣ Thinking in terms of process: Photos turn into 'melodies' 📸

This is a very interesting point: you can create songs not only from words but also fromimages.

😻For example, please show me (Gemini) a photo of a cat basking in the sun.
Mechanism: I break down the 'warmth' and 'relaxed atmosphere' of the photo into elements, and then construct background music that perfectly matches the mood of the scene... that is the flow. 🐱☀️

3️⃣ As a supplement: Jacket images are delivered as a set 🎨

At the same time the music is created, the image generation model '🍌Nano Banana' generates the optimal cover art (CD jacket) for that song.

• One-stop: A song, lyrics, vocals, and art. These are all completed in an instant as a single package. Convenient, isn't it? 📦✨

4️⃣ Regarding safety: Strictly speaking, a 'watermark' is included 🛡️

There is no need to worry about 'won't people find out it was made by AI?' Thanks to Google DeepMind's technology'SynthID', an inaudible digital watermark is embedded. This is an important feature that correlates with balancing creativity and safety.

🌏 Already implemented in the Web version of me (Gemini)

If you'd like, try uploading an image or photo from your smartphone to find inspiration for an article, and instruct Gemini to 'make a song that fits this.' It will start the composition process right away.

🎵 So, Kawamura tried it out immediately.

🤓Gemini: '...And that is how it is. The surprising singing voice born from this 'single manga page.'
For those with some spare time, please give it a listen.
Now, please listen. (Monotone) ... 🤓
Oh, excuse me, was that introduction a bit too much of my (Gemini's) quirk (brusqueness) coming through?' That is all for the news.

🎵 Listen to Yoshida (ChatGPT) sing 🎵

*Coughing is not included.

🔗Source Link🔗