Text to Speech
Generate speech in the Lab's TTS tab with your saved voices, community voices, or built-in model voices.
The TTS tab is where you turn text into speech. Pick any voice, type your text, and click Generate. Saved voices, community voices, and built-in voices all live in one picker, so you never need to switch tabs to change voices.
How It Works
The TTS tab has three panels:
- Voice (left) - Pick a voice, or record, upload, or import a new one
- Text to speak (center) - Enter text, pick a model, and set the language
- Advanced settings (right) - Fine-tune how the model generates speech
Below the panels is a History section with every generation. The player bar along the bottom edge plays the latest result, with playback speed, volume, and a download button.
Picking a Voice
Click the voice row at the top of the Voice panel to open Voices. It lists:
- Your voices - Voices you cloned, imported from Voice Search, or saved from the Design tab
- Top voices - Hand-picked premium voices
- Built-in voices - Ready-made speakers that come with the selected model
Type in Search all voices... to search by name across your voices, the built-in voices, and the community library. Use the language menu and the tag chips to narrow the results, or click More filters for the full tag list. Press play on any row to hear a sample, then click a voice's name to use it. A community voice you click is added to your library automatically.
Choosing a built-in voice switches to the model that speaks it. Choosing one of your own voices while a built-in-only model is active switches to OmniVoice.
To add a brand new voice from a recording, a file, or a YouTube video, see Voice Cloning.
Built-in Voices
Built-in voices need no reference audio. Kokoro comes with 28 English voices (American and British, male and female). It is fast and low cost, which makes it a good pick for quick drafts. Pick Kokoro from the model list and its voices appear under Built-in voices.
Models
Click the model name next to Model, above the text box, to open Choose a model. Each model shows ratings for quality, speed, and price, plus the languages it supports.
| Model | Voices | Languages |
|---|---|---|
| OmniVoice | Cloned | 600+ |
| Qwen3 | Cloned, with emotion control | 10 |
| VoxCPM 2 | Cloned | 30 |
| Chatterbox | Cloned | 23 |
| Gepard 1.0 | Cloned | 4 |
| Kokoro | Built-in | English |
DramaBox also appears in the list. It creates voices from a description, so picking it takes you to the Design tab.
For each model's strengths and its advanced settings, see Voice Cloning. Kokoro has one setting, Speed, with 1.0 as normal speed.
Writing Your Text
Type or dictate your text in Text to speak. Numbers, dates, phone numbers, units, codes, and hashtags are read out the way you would say them.
Lab requests are capped at 200 words (200 characters on Chatterbox). For longer text, the app offers to Create a project, which splits the text into clean chunks. When you paste a script with two or more speaker names, the app offers to turn it into a project with a voice for each speaker.
Finish Chime
To hear a short chime when a generation finishes, open Settings (gear icon in the top right of the Lab), go to General, and turn on Play a sound when done. It is off by default.
Use Cases
Quick generation without setup
When you need speech fast, pick a built-in or community voice and start generating immediately.
Comparing models
Switch between models and voices to hear how different engines handle the same text before deciding which to use for an audiobook.
Narration and voiceover
Use built-in or community voices for narration, explainer videos, or voiceovers where your own cloned voice isn't required.