Studio · Text to speech
Text to speech that sounds directed.
Most text to speech reads a block of text. Leelo reads a script: every speaker their own voice, every line directed, sound on its own lanes, one finished file at the end.
01How it works
Three steps, start to finish.
- Step 1
Paste your script
Leelo splits it into lines and finds the speakers. Or start from a template: podcast, YouTube, audiobook, course, ad, phone menu.
- Step 2
Cast and direct
Pick a voice for each speaker, then set the pace, emotion and emphasis of any line. Hear each one as you go.
- Step 3
Mix and export
Add sound effects and music on their own lanes, set the levels, and export one file at the loudness your platform wants.
02Made for
Where people use it.
- YouTube & videoVoiceovers paced to your cut“Three years ago, this valley was underwater.”A 10-minute voiceover ≈ 1,500 credits
- PodcastsA host, a guest, a whole show“Welcome back to Field Notes. Today: why cities are getting quieter.”A 30-minute episode ≈ 4,500 credits
- AudiobooksChapters, one steady narrator“Nobody had lived in the house for eleven years.”An 80,000-word novel ≈ 80,000 credits
- E-learning & trainingLessons people finish“In this lesson, we'll look at how compound interest works.”A 10-module course ≈ 7,500 credits
- Phone menus & IVRGreetings that sound like you“Thank you for calling Northgate Dental.”A full phone menu ≈ 200 credits
03Details
Control where it matters.
The read you'd ask a voice actor for, without the booth.
- Voices
- Hundreds of voices in most languages, many with emotions like cheerful, sad or whispering.
- Direction
- Pace, pauses and emphasis per line, on the voices that support them.
- Pronunciation
- Teach it a name or a brand once, and every project says it right.
- Long form
- Whole books, chapter by chapter, with chapter marks in the export.
- Export
- MP3, WAV or M4B (audiobooks), at −16 LUFS for podcasts, −20 for audiobooks, or −23 for broadcast.
Did you hear that? Someone's at the door.
Pace
0.9×
Emotion
Whispering
Emphasis
“door”
04Questions
Before you start.
Text to speech has moved on from a single voice reading a block of text. This guide explains how Leelo's Studio works, the controls that make AI speech sound directed, and how to get finished audio out the other end.
Scripts, lines and speakers
Everything in the Studio is a script made of lines. Paste text and each paragraph becomes a line; give a line a speaker and it takes that speaker's voice. A project can have as many speakers as it needs, from a single narrator to a full cast.
Templates give you a head start for common jobs: YouTube voiceovers, podcasts, audiobooks, e-learning, ads and phone menus. For a single block of text, quick text to speech opens a one-voice project straight away.
Choosing voices
The voice library has hundreds of voices in most of the world's languages, with accents within them, and every voice has a sample to play. Standard voices cost a credit a word, premium voices 1.5, and your own cloned or designed voices 5.
Many voices can perform emotions such as cheerful, sad, angry or whispering; multilingual voices can read several languages in the same voice.
Directing a line
Direction is what separates a voiceover from a read-through. Per line, you can set the pace, pick an emotion on voices that have them, add pauses and emphasise words on voices that support it. Because every line is generated on its own, you can redo one line until it's right without touching the rest.
- Pace: slow a line down for weight, speed it up for energy.
- Pauses: give key facts and scene changes room.
- Emphasis: put the stress where the meaning is.
- Pronunciation: teach names and brands once, for every project.
Sound, music and the mix
Below the script is a timeline with a lane per voice. Sound effects you describe go on lanes of their own, as does music, each with a level you can hear as you set it. What you export is that mix.
Exporting
Export MP3 or WAV, and M4B for audiobooks, at the loudness your platform expects: −16 LUFS for podcasts, −20 for audiobooks, −23 for broadcast, or untouched. You can also export stems, one file per voice and all the same length, for mixing elsewhere.
Free and paid
The free plan gives you 1,000 credits once, when you sign up: about 1,000 words, roughly seven minutes of narration in a standard voice, and every tool. Paid plans add credits every month, commercial use, more storage and files kept for longer. See pricing for the details.
Write it. Voice it. Ship it.
1,000 free credits to try everything. No card needed, and nothing to install.
Start free: 1,000 credits