ElevenLabs V3 turns a script into speech with a selectable voice, language guidance and delivery cues in the text. Choose from 21 voices, use audio tags to express a whisper, laugh or change in emotion, and adjust stability while auditioning the result. Text normalization helps prepare numbers and similar written forms for speech, while optional timestamps return timing information alongside the audio. Prepare a complete script, choose its voice settings and review the character-based estimate before generating narration, dialogue or educational audio.
ElevenLabs V3 Text to Speech is developed by ElevenLabs. Vidgo provides access to the model. Model names and trademarks belong to their respective owners.Open the V3 task with the words to speak, select a supported voice and optionally specify the language. Add meaningful delivery tags where the narration calls for a change, then choose stability, text normalization and timestamps. Open the task to audition the voice and prepare API integration.
The text field contains the script and any inline audio tags; voice selects the speaking voice. stability, language_code and apply_text_normalization guide the delivery and text processing, while timestamps requests timing output. Keep the original spoken language when comparing a record with its audio; page translations explain the meaning of the source text.
| Model / task | Input and workflow |
|---|---|
| ElevenLabs V3 Text to Speech | Expressive text to speech with 21 selectable voices, audio tags, language control, and optional timestamps. |
Existing recordings include an English attic discovery, a Mandarin family memory and an English science explanation with timestamps requested. Listen to the different voices and inspect the original text and settings to understand the controls in use. Each is a separate script and recording, with its own language and creative purpose.
Use inline tags such as [whispers], [curious], [excited] and [laughs] with the spoken words. The attic record moves from quiet discovery to surprise and relief; listen for how the selected voice interprets the transitions and pauses.
Supply the script in its intended language and optionally set language_code, then choose a supported voice. The Mandarin dumpling record uses Rachel and zh, giving you a source for reviewing word clarity, conversational rhythm and expressive cues in Chinese.
Set timestamps true when the application needs the returned timing data and read all entries in files. The science record also uses normalization on for a script containing 440 Hz, so you can listen to number pronunciation while retrieving its separate timestamp attachment through the task response.
[whispers] Wait... that ticking is coming from inside the wall. [curious] This little brass key fits the crack. One turn, and... [excited] the whole ceiling is full of stars! [laughs] Grandpa, you built a planetarium in the attic. [exhales] All these years, it was waiting for us.
Place cues next to the words whose delivery should change and use punctuation to express pauses. The displayed attic input is the original English text; its whispered opening, discovery and release provide a complete script to compare with the recording.
{
"voice": "Rachel",
"language_code": "zh",
"stability": 0.5,
"timestamps": false,
"apply_text_normalization": "auto"
}Select one of the 21 supported voices, adjust stability from 0 to 1 and choose normalization auto, on or off. Keep those settings with the script when comparing takes. The displayed Mandarin configuration records the actual Rachel voice, zh language and normalization choice used for that audio.
The attic request uses Charlotte with English delivery tags to tell a discovery story. The dumpling request uses Rachel with Mandarin text and expressive cues for a family memory. Read the original scripts, voice configurations and meaning translations while listening to each corresponding result. Compare words, pauses, emotional changes and the chosen voice within each record.
[whispers] Wait... that ticking is coming from inside the wall. [curious] This little brass key fits the crack. One turn, and... [excited] the whole ceiling is full of stars! [laughs] Grandpa, you built a planetarium in the attic. [exhales] All these years, it was waiting for us.
{
"voice": "Charlotte",
"language_code": "en",
"stability": 0.5,
"timestamps": false,
"apply_text_normalization": "auto"
}Original Mandarin spoken text: [sighs] 我练了一个下午，面团还是歪歪扭扭的。奶奶看了一眼，说：别急，手放轻一点。[curious] 这样吗？先对折，再慢慢捏紧……[laughs] 真的站住了！虽然像只胖企鹅，但这是我第一次包好饺子。原来，有些本事不是听懂的，是一起做着学会的。 English meaning: [sighs] I practiced all afternoon, but the dough was still crooked. Grandma took a look and said, “Take your time. Use a lighter touch.” [curious] Like this? Fold it in half first, then pinch it closed slowly... [laughs] It really stands up! It may look like a chubby penguin, but this is the first dumpling I have ever made properly. Some skills, it turns out, are learned by doing them together rather than just understanding an explanation.
{
"voice": "Rachel",
"language_code": "zh",
"stability": 0.5,
"timestamps": false,
"apply_text_normalization": "auto"
}Prepare a complete script, choose a speaking voice and describe the delivery through punctuation and appropriate tags. Generate and listen to the result before using it as a story narration, character line, language example, educational explanation, guided experience or application speech response. Request timestamps when your workflow also needs timing data.
Write a scene with clear emotional transitions and pauses, then choose a voice and delivery tags. Listen to the complete recording for continuity, word clarity and atmosphere before placing it in an audio story.
Prepare a short line and its emotional intent, then add focused delivery cues and select a voice. Audition the generated speech to develop a character moment or dialogue prototype.
Write a concise concept explanation with the numbers and terms that learners need. Choose normalization and optional timestamps, then review pronunciation and pacing for a lesson or demonstration.
Prepare text in the intended language and set its language code when appropriate. Generate and review the spoken words, pauses and expression as an audio example for a learning activity.
Write clear instructions with a reassuring delivery and deliberate pauses. Audition the voice for comprehensibility and rhythm before using the result in a tour, onboarding sequence or guided activity.
Collect the script and selected voice settings, submit the documented request and retrieve the audio by task ID. Preserve timing attachments when requested so the application can use speech and its associated data together.
V3 is billed by the actual number of Unicode characters after leading and trailing whitespace is trimmed, using the current rate per 1,000 characters. Calculate the request amount proportionally from the script length. The dynamic table shows the current character rate; open the task with the complete text to review its estimate, and budget additional takes using each submitted script’s length.
| Model | Provider | Workflow | Input material | Speech output | Starting rate |
|---|---|---|---|---|---|
| ElevenLabs V3 Text to Speech | ElevenLabs | Text to Speech | Spoken text and voice settings | Speech / optional timestamps | $0.08 / 1,000 characters |
Test your creative brief online, then bring the verified request into your application.
Open the V3 task, enter the full script and select voice, language and text-processing settings. Review the character-based estimate and generate. Listen through the complete recording for pronunciation, pauses and emotion; when timestamps are requested, retrieve their separate file together with the audio.
Choose a task, inspect its settings and cost, and follow the result from an online test to an authenticated API request.
Prepare the words in the language to be spoken and place appropriate delivery tags in context. Choose a supported voice, stability and normalization, then request timestamps when needed. Keep the original text with the settings so you can compare the actual delivery across takes.
Open the V3 task, enter the full script and select voice, language and text-processing settings. Review the character-based estimate and generate. Listen through the complete recording for pronunciation, pauses and emotion; when timestamps are requested, retrieve their separate file together with the audio.
Use the trimmed script’s actual Unicode character count and the current per-1,000-character rate. Plan the number of revised takes using each version’s full text and review the task estimate before generating.
Read the original spoken text and voice settings while listening to the result. Translations explain the meaning while preserving the recording’s original spoken language. Review pronunciation, pacing and expression before adapting the request to your own script.
Retain the task ID returned by the API to query status and retrieve completed file URLs. Use your online generation history to review earlier results and the creative direction you tested.
Use the documentation linked from each task for its exact model identifier, accepted inputs, authentication and response format. Integrate the same task you tested online so its media and settings match the request.
ElevenLabs V3 is a text-to-speech model on Vidgo with 21 supported voices, inline audio tags, language guidance, stability control and optional timestamps. It accepts up to 5,000 Unicode characters after trimming and returns speech audio with separate timing files when requested.
Read speech-model introductions, API guides and platform comparisons to understand script preparation, voice controls, integration and character-based costs.
