Get started · Lesson 6 of 7
How to make AI voices, sound effects and music
A narrator, a full audio scene with voices and effects, and an original score, each from one prompt.
You need: The words you want spoken, or a scene to describe.
Step by step
Open Seed Audio
One box makes speech, sound effects and whole scenes. Type the words you want spoken, or describe a scene.
Describe a scene for a full mix
"Rain on the window, thunder, a tired detective, a nervous client, a door creaking open." Seed Audio voices the characters and lays the effects under them.
Put narration in quotes
Describe the voice, then quote the script: a warm British narrator for a nature documentary, "Every autumn, the red deer return to the misty valley." Quotes tell it exactly what to say.
Tune speed, volume and pitch
The controls sit under the prompt. Speech costs about a tenth of a credit per second.
Score it in Text to Music
Describe the style, mood, instruments and whether you want vocals, then pick a length from 30 seconds to five minutes. The price updates as you go.
Why it works
- Describe the speaker before the line. Age, accent and mood shape the read more than punctuation does.
- Score last, after the cut. Music written to the final length lands its ending on time.
Your practice task
Write a 15-second radio ad: one narrator line in quotes, one sound effect, then a 30-second bed in Text to Music.
Read the transcript(2:18)
Great video needs great sound. Today, a narrator, a full audio scene with voices and effects, and an original score, all from a prompt. Open Seed Audio. One box does it all. Type the words you want spoken or describe a whole scene. Let's make a tiny radio drama. Rain on the window, thunder, a tired detective, a nervous client, and a door creaking open. Click generate. A few seconds later, here it is. You're late. I had to be sure I wasn't followed. For narration, describe the voice you want, then put the script in quotes. A warm British narrator for a nature documentary. Every autumn, the red deer return to the misty valley, and for a few short weeks, the whole forest listens. Speed, volume, pitch, and format are all down here, and it costs about a tenth of a credit per second. Now the music. Open text to music and describe the track, the style, the mood, the instruments, and whether you want vocals. Pick a length from 30 seconds up to five minutes. The price updates as you go. Then click generate. Under a minute later, you have an original score ready to download. Every autumn, the red deer return to the misty valley, and for a few short weeks, the whole forest listens. Put the three together, and that's a soundtrack. See you in the next tutorial.