AI text to speech
Hear your script in your saved voice
Turn a script into speech using a voice you have cloned and saved in your account. Choose the voice, language and delivery controls, then preview the generated reading. Download the audio or use it in a talking video. This workflow starts with your own saved clones; it has no public catalog of stock speakers.
Available models, settings and costs are shown inside the product before you create.
The first time I opened this notebook, every page was blank. A month later it held sketches, shopping lists and the beginning of a story I wanted to finish.
How it works
Choose a voice and shape the reading
- 01
Select a voice from your collection
Open My Voices and choose the speaker for the piece. If you have not created a voice yet, add a clear, authorized recording through the voice creation form. Give the clone a recognizable name so you can find the same voice for later scripts and videos.
- 02
Write the script and choose the delivery
Enter the words you want spoken and select a supported language. Adjust the speed in any supported language; emotion controls are available in English only. Read the text aloud once before generating to catch awkward phrasing, long sentences or a number that needs to be written more clearly.
- 03
Generate, listen and keep the reading
Review the credit estimate for your text and create the speech. Listen for the opening, sentence endings and any names or specialized terms. You can download the generated audio, or use the reading with a still photo in Lip Sync or a video in the narration flow.
Practical guidance
Write for a listener, then check the reading
Text that looks tidy on a page can sound crowded when spoken. Give each sentence one clear job and leave room for the listener to follow the idea. The preview lets you hear where the script needs another pass.
Put the useful information early
For a tutorial, start with what the viewer is about to learn or do. For a product explanation, name the problem and the relevant action. Spoken introductions get tiring when the listener has to wait through several sentences before discovering what the piece is about.
Write names and numbers for the ear
Test any unfamiliar name before generating a full passage. A date, a price or an abbreviation can be read in more than one way, so write the intended pronunciation in ordinary words where needed. Listen again after changing it; spelling alone does not prove how the voice will say it.
Use the delivery controls deliberately
Begin with a normal pace and listen before changing it. Speed choices work in supported languages, while emotion controls are available in English only. For English speech, try a calm or excited reading when it fits the script. Compare the previews, since a setting guides the performance without guaranteeing an exact tone.
Prepare the script in the intended language
Speech supports multiple languages, including English and Hebrew. Supply the words you want spoken in that language and choose it in the composer. For a multilingual project, check pronunciation and natural wording in each version. Language selection does not write or translate the script for you.
Finish the audio before creating the video
A pause or a revised line changes the timing of a spoken piece. Settle on the reading before you make a talking photo or mix narration into a clip. That gives you a known audio track when you judge mouth movement, visual pacing or the length of the final video.
Questions
Frequently asked questions
Can I choose a ready-made public voice?
This speech workflow uses voices you have cloned and saved in your own account. Create a voice in My Voices before generating a reading. The page lists your saved voices, with controls to try a script, manage the voice and continue into a video.
Does text to speech support Hebrew?
Yes. Hebrew and English are among the supported speech languages. Choose the language of your script in the composer and listen to the generated reading. Check names, borrowed words and mixed-language sentences carefully before using them in a finished piece.
Can I change how quickly the voice speaks?
The speech composer offers slow, normal and fast delivery choices in supported languages. Emotion controls are available in English only. Listen to a short passage at the setting you want, since the exact duration still depends on the text and how the voice reads it.
How is the speech cost shown?
The composer displays a credit estimate based on the text in the request. Check it after making your final edits and before generating. If you use that speech to create a talking video, the video has a separate cost for its selected setup.
Start with the idea you have now
Open My Voices, select a voice you have created and try a short script. Review the speech estimate before generating, then listen before using the audio.
Create speech