Runway and ElevenLabs
Level 5 · Pictures, video and voice
Runway turns words or a photo into a short video clip, and ElevenLabs turns text into a human voice. What each can do, and how to use them fairly.
Course contentsLesson 10 of 10
Level 1 · Chat and answer apps
Level 2 · AI at work
Level 3 · AI in the code editor
Level 4 · Agents and app builders
Level 5 · Pictures, video and voice
After this lesson you can
- Say what Runway and ElevenLabs each make
- Explain why clips and copied voices need care
- Use a real face or voice only with consent
11 mincovers 6 glossary entries
Worth reading first: Midjourney and AI Image Tools
How AI makes video clips and voices
Runway and ElevenLabs make two things a picture tool cannot. One makes moving video, and the other makes a voice that sounds human. Runway, from New York, turns a line of text or a photo into a clip a few seconds long. ElevenLabs began in 2022. It turns text into speech, and one of its voice models speaks 74 languages.
Runway does text-to-video. Its video models are named Runway Gen, with a number for each new version. You give it a prompt or a photo, and it makes a short clip. Its editing tools can also remove a background, change the look of a scene or make a shot run longer.
Clips are short, so a longer scene is built from many clips joined together. Faces, hands and written signs can shift from one frame to the next. Film makers and ad makers use it to try an idea in minutes, before they pay for a real shoot.
ElevenLabs does text-to-speech. You paste text, pick a voice and get speech with pauses, stress and feeling that sound human. You can choose from a library of voices or design a new one. Apps can use the same voices through its programming interface, for audiobooks and phone helpers.
It can also copy a real voice from a recording. This is voice cloning, and it is the part to handle with care. A copied voice can say anything, including things the real person never said. Use it only with the owner's clear permission, and never to fool someone.
Its dubbing tool takes a video in one language and gives it back in another. It keeps the speaker's own voice. A teacher can record a lesson once and share it in Spanish, Hindi and Arabic. Someone who speaks each language should still check it. A wrong word sounds just as sure as a right one.
Put the two together, and you can make a person seem to say something they never said. That is a deepfake. The tools add marks and rules to slow misuse. The best guard is your own habit: label AI video and voice as AI, and get consent before you use a real face or voice.
Which one to pick
- 1
Pick Runway
When the job is a moving picture. A short shot for an ad, a mood clip, or a test before a real shoot.
- 2
Pick ElevenLabs
When the job is a voice. A book read out loud, a voice-over, or a video in a new language.
Worked example
Emma makes a bakery ad
Emma needs a 30-second ad for her bakery. Runway turns three photos of her bread into three short moving shots, and she joins them in an editor.
ElevenLabs reads her 60-word script in a warm voice. It says the bakery's name wrong twice, so she spells it out the way it sounds, and the third try is right.
The words for this lesson6
Check yourself
When is it fine to clone someone's voice with ElevenLabs?
Finished this lesson?
Marking it complete also marks its 6 glossary entries learned, which updates your roadmap and your saved page. 0 are already marked.