The [Generate audio] app does three things with AI voices: reads a script aloud, replaces the voice in a clip, and translates speech into another language. Results land on your timeline. Generating uses Selects credits.
What's in the Generate audio app
The Generate audio app has three tabs.
Tab | Use it to |
[Text to speech] | Turn a script into a voice |
[Voice change] | Replace the voice in a clip and keep the delivery |
[Translate] | Translate and re-voice the speech in a clip |
Turn a script into a voice
To turn a script into a voice, use the [Text to speech] tab.
Open the draft you want the audio added to.
In the project panel, click the [Apps] tab, then click [Generate audio].
Choose a [Model].
Write exactly what the voice should say under [Script].
Choose a [Voice]. With ElevenLabs voices, click the play button to hear a sample first.
Optional: set [Takes] from 1 to 4 to get several reads of the same script.
Click [Generate].
The voice speaks the language your script is written in. There's no language picker. Seed Audio 1.0 has a [Multilingual] switch under [Advanced settings].
Note: Each take is its own generation and is charged separately.
Model | Good for | Script limit |
ElevenLabs v3 | Natural, stable narration | 5,000 characters |
MiniMax Speech 2.8 HD | Controlling mood, speed, and pitch | 10,000 characters |
Seed Audio 1.0 | Expressive speech that can match a voice from a clip | 2,048 characters |
Tip: To match a real voice, choose Seed Audio 1.0 and add up to 3 clips of that voice under [Add references], each up to 30 seconds. The reference replaces the preset voice.
Change the voice in a clip
To change the voice in a clip, click the [Voice change] tab. Add your clip under [Your clip], pick a [Voice], and click [Generate]. The new voice keeps the original timing and delivery. Turn on [Remove background noise] to clean up the recording at the same time.
The clip can be audio, or the sound of a video clip, up to 5 minutes long.
Translate the speech in a clip
To translate speech, click the [Translate] tab. Add your clip under [Your clip], choose the [Language] to translate into, and click [Generate]. Selects translates the speech and re-voices it. The clip can be up to 5 minutes long, and there are 29 languages to choose from.
Where your audio lands
Audio generates in the background. When it's done, Selects saves it to your computer's Downloads folder, adds it to your project, and places it below your other audio, at the spot where the playhead was when you clicked [Generate].
Caution: Your project reads the audio from the file in your Downloads folder. If you move or delete that file, the clip goes offline.
How credits work
When the model can be priced in advance, the credit cost shows just above [Generate]. If no estimate shows, the credits charged appear on the result under [Recent] once it's done. If you don't have enough credits, click [Get credits] to open your plan page.
Your last 3 audio generations in this project show under [Recent], with [Try again].
FAQs
Can Selects generate music or sound effects?
The Generate audio app doesn't generate music or sound effects. For ready-made sound effects, open the [Effects] app and click the [Sounds] tab.
Can I clone my own voice?
You can't save a cloned voice in Selects. To get close, choose the Seed Audio 1.0 model in [Text to speech] and add up to 3 clips of your voice as references. Selects matches that voice for the script you write.
What languages can I translate speech into?
You can translate speech into 29 languages, including English, Korean, Japanese, Chinese, Spanish, French, German, Portuguese, Hindi, and Arabic. Choose one under [Language] in the [Translate] tab.
Where did my generated audio go?
Your generated audio is on your draft's timeline, below your other audio, at the spot where the playhead was when you clicked [Generate]. It's also in your project panel and your computer's Downloads folder.