AI Creative Tools: Image, Video, Music and Voice Generators Explained

A few years ago, making a picture meant drawing it, painting it or taking a photo. Now you can type a sentence and get one back in seconds. The same goes for video clips, songs and voices.

The Creative Tools group is the bottom row of the Periodic Table of AI, numbers 89 through 103, in purple. These are the 15 best-known tools for making images, video, music and speech. This post walks through all of them, then covers the part every creator needs to know: who owns what you make.

A glowing purple sphere sends streams of light that turn into a paintbrush, a strip of film, a music note and a sound wave
One prompt, four kinds of output: pictures, video, music and voice.

The last post covered everyday AI applications. This one is about the creative side.

Most of these tools run on the diffusion model idea from the Core Concepts post. They start from random noise and clean it up, step by step, until it matches your description.

Images

Midjourney

Midjourney is known for striking, artistic images, and it’s popular with designers and illustrators. It also landed in one of the first big AI art controversies.

In 2022, Jason M. Allen won the digital art category at the Colorado State Fair with a piece called “Théâtre D’opéra Spatial,” made with Midjourney. He disclosed that he’d used it, but the win still set off a fierce debate. When he applied for copyright, the U.S. Copyright Office turned him down, and in 2023 its review board made that final, finding the work didn’t have enough human authorship. Allen took the fight to federal court in 2024.

OpenAI Images

OpenAI Images started with DALL-E, which OpenAI revealed in January 2021, one of the first systems to turn a sentence into a picture. In March 2025, OpenAI replaced DALL-E inside ChatGPT with image generation built right into its GPT-4o model. Now you can create and edit images just by chatting.

Stable Diffusion

Stable Diffusion is the open-weight option that changed everything. Stability AI released it on August 22, 2022, together with researchers from LMU Munich and Runway. Because anyone could download it and run it on a home computer, it set off a huge wave of community tools, custom styles and add-ons.

Flux

Flux comes from Black Forest Labs, a German company founded in 2024 by former Stability AI researchers whose earlier work led to Stable Diffusion. They released the first Flux models in August 2024, some with open weights. It’s known for following prompts closely.

Ideogram

Ideogram made its name on something image generators used to be terrible at: text. Getting a readable word onto a sign, poster or logo used to be nearly impossible for AI. Ideogram focused on getting lettering right, which makes it handy for designs that need words in them.

Adobe Firefly

Adobe Firefly is Adobe’s family of creative AI models, introduced in March 2023 and built into Photoshop, Illustrator and Adobe’s other apps. Its big selling point is where its training data came from. Adobe says its first model was trained on Adobe Stock images, openly licensed content and public domain material whose copyright has expired. That pitch is aimed squarely at businesses worried about legal risk.

Video

Runway

Runway is a creative AI company that helped build the foundation of this whole field. It was one of the partners behind the original Stable Diffusion release. Today, it’s best known for AI video generation and editing tools used by filmmakers and video creators.

Sora

Sora is OpenAI’s video model, and its story shows how fast this field moves. OpenAI unveiled it in early 2024, and the clips stunned people. In 2025, OpenAI launched Sora 2 with its own app, which shot to the top of Apple’s App Store. Then, in March 2026, OpenAI announced it was shutting down the Sora app and its developer service, citing computing costs and a shift in priorities. Sora is still on the chart because it changed what people thought AI video could do.

Veo

Veo is Google DeepMind’s video model. Its big leap came with Veo 3 in May 2025, which could create sound to go with its clips: sound effects, background noise and even dialogue. Until then, most AI video was silent. Google also marks Veo videos with an invisible watermark called SynthID to help identify them as AI-made.

Kling

Kling comes from Kuaishou, one of China’s biggest short-video platforms. It launched in June 2024, and it was one of the first text-to-video models the public could actually try. Kuaishou’s huge library of short videos gave it a big head start.

Music and voice

Suno

Suno makes complete songs, with vocals, lyrics and instruments, from a short text description. You describe the style and the mood, and it writes and performs the track.

Udio

Udio does the same kind of thing, and the two companies have shared a big legal story. In 2024, the major record labels, Universal, Sony and Warner, sued both Suno and Udio, claiming they trained their AI on copyrighted recordings without permission. The startups argued it was fair use. In late 2025, the fight started turning into deals. Udio settled with Universal and Warner, and Warner settled with Suno, with plans for new licensed AI models.

ElevenLabs

ElevenLabs is a leading voice AI company. It can turn text into natural-sounding speech, translate a voice into other languages, and clone a voice from a sample. Audiobook narration, video voiceovers and accessibility tools all use this kind of technology.

Voice cloning also has an obvious dark side. In January 2024, voters in New Hampshire got robocalls with an AI-cloned voice of President Biden telling them not to vote in the primary. The political consultant behind it faced criminal charges and a proposed $6 million fine from the FCC, which ruled that AI-generated voices in robocalls are illegal.

HeyGen

HeyGen makes videos of AI avatars, digital presenters that speak whatever script you give them. Businesses use it for training videos, marketing and translating videos into other languages while matching the lip movements.

All-in-one

Canva AI

Canva AI brings AI image generation, writing help and design tools into Canva, the popular online design app. Its appeal is convenience. Instead of jumping between separate AI tools, you can generate and edit everything inside the app where you’re already making the social post, presentation or flyer.

What creators should know about ownership

Here’s the part people skip. In the United States, the Copyright Office released a report in January 2025 that says copyright requires a human author. Material generated entirely by AI isn’t protected, and the Office said that typing prompts alone doesn’t make you the author.

That doesn’t mean AI-assisted work can’t be protected. The Office said your own creative contributions can be, such as human-made material that shows up in the final piece, or a creative selection, arrangement or editing of the AI’s output. The more of you there is in the work, the stronger your claim.

Training data is a separate fight that’s still playing out, as the music lawsuits show. If you’re using AI images, music or voices for a business, read each tool’s license terms and lean toward tools that are clear about their training data and commercial use. This isn’t legal advice, and the rules are still changing, so check the latest before you build a business on it.

Where to go next

Creative tools raise some of the hardest questions in AI: about ownership, deepfakes and trust. That leads straight into the next group on the chart, Safety & Ethics, and that’s the next post.

For now, click any purple tile and read the full page. Start with Stable Diffusion if you want to understand how this whole field got going.

1 thought on “AI Creative Tools: Image, Video, Music and Voice Generators Explained”

Comments are closed.