# Text to speech: turn a script into a natural voiceover

Pick Eleven v4 or Gemini 3.8, choose a voice, direct the delivery, and download the MP3.

## How it works

1. Type or paste your script.
2. Choose a model and a voice. For Gemini, add a style such as "warm and calm".
3. Generate the speech, play it, and download the MP3.

## A voiceover from a script, in the voice you choose

Audo turns a written script into an MP3 voiceover with four voice models side by side: Eleven v4 and Eleven v4 Turbo from ElevenLabs, and Gemini 3.8 Flash and Flash-Lite from Google. Pick the model, pick one of its voices, and Audo reads up to 8,000 characters in one go, about eight minutes of speech. Each model keeps its own way of being directed, so the tags you already know work as they do on the model's own site.

The models are good at different things. On Artificial Analysis' blind listening test, people preferred Eleven v4 over every other model. Gemini 3.8 Flash speaks more than 130 languages and costs the least, and listeners on Design Arena preferred it when they described the voice they wanted.

Use the MP3 under a video, as a podcast intro, for a course lesson, or for a scene with several characters, one line at a time. Mix it with music in Mix audio, or set it to a waveform in MP3 to MP4. The same tool works from your AI assistant through Audo's MCP server, and from code through the API.

Tool: Text to speech (`text_to_speech`). Price: 20 credits per 1,000 characters — $0.20 per 1,000 characters, with no subscription — per job’s measured quantity, charged only when a job succeeds.

## Free API

```bash
curl -F 'text=Welcome back to the show.' -F model=eleven_v4 https://audo.ai/api/try/text-to-speech -D response.headers -o response.body
```

Read response.headers and response.body before naming media. Handle 202, 200 JSON and errors with the complete response client at https://audo.ai/docs/api?utm_source=markdown&utm_medium=agent#responses.

Free for the first 300 source characters, ending at a complete cue, into up to undefined languages, 3 times a day per person. SRT or VTT uploads up to 2 MB. Free results last one hour. The response is the result itself; a job that runs longer than about 30 seconds returns 202 with a link to check on it. With an API key (`-H "Authorization: Bearer $AUDO_API_KEY"`), omitting free runs the whole file with credits if the balance covers it, otherwise the free portion if quota remains. Set free=true to prohibit spending; for paid whole files use free=false with an approved positive max_credits. A free result cut short comes with finishUrl (and the Audo-Finish-Url header when the response is the file itself): give it to the person, who pays there (from $5, no sign-up), and the whole file runs; keep checking the status link, whose finishedBy names the whole-file job and its own status link. Details: https://audo.ai/docs/api.md

## Options

Form fields with the same names as the MCP tool's inputs:

- `text`: The script to read, up to 8,000 characters, such as "Welcome back to the show." Pass text or text_file_id, not both.
- `text_file_id`: 
- `model`: The voice model and its price per 1,000 characters: eleven_v4 (Eleven v4, 20 credits), eleven_v4_turbo (Eleven v4 Turbo, 10 credits), gemini_3_8_flash (Gemini 3.8 Flash TTS, 8 credits), gemini_3_8_flash_lite (Gemini 3.8 Flash-Lite TTS, 5 credits). Default eleven_v4.
- `voice`: The model's voice. Eleven: george, sarah, brian, alice, daniel, lily, adam, matilda (default george). Gemini: one of its 30 voices, such as kore (default), puck, or charon.
- `style`: Gemini models only: how to deliver the whole script, up to 2,000 characters, such as "warm and calm" or a paragraph describing the speaker and pace.
- `language`: The script's language as a code, such as es or ja. Default: detected from the text. Every model detects it; naming it limits the job to models that speak it and helps Eleven read numbers and dates.

## Languages

Any language: it works on the sound, not the words.

## From an assistant (MCP)

Add https://audo.ai/mcp and sign in (setup for each assistant: https://audo.ai/setup.md). The tool is `text_to_speech`:

> Reads a script aloud as an MP3, up to 8,000 characters, with one model and voice: eleven_v4, eleven_v4_turbo, gemini_3_8_flash, gemini_3_8_flash_lite (default eleven_v4). Direct Eleven with [tags] in the text, and Gemini with style. It costs 5 to 20 credits per 1,000 characters by model, at least 1, charged only if the job succeeds; free up to 300 characters, 3 times a day. Returns the audio with download links, or job_id for get_job if it takes longer.

Example: "Read this script aloud with Eleven v4 in the George voice."

## Questions

**Is Audo's text to speech free?** Audo reads scripts up to 300 characters free, about 20 seconds of speech, 3 times a day, with any model and no sign-up. Longer scripts, up to 8,000 characters, are paid per 1,000 characters at each model's rate, shown before you generate, and charged only if it works. There is no subscription.

**Which model should I pick?** Eleven v4 ranked first on Artificial Analysis' blind listening test with each model's own voices, and Eleven v4 Turbo is close behind at half the price. Gemini 3.8 Flash ranked first on Design Arena, where listeners steer the voice with a description, and costs the least.

**How do I direct the voice?** Eleven v4 and Eleven v4 Turbo follow directions in square brackets inside the script, such as [whispers] or [laughs]. Gemini takes a style for the whole script, such as "whispered urgently", and short tags in angle brackets, such as <laugh>.

**How much does a long script cost?** Eleven v4 is $0.20 per 1,000 characters, about a minute of speech; Eleven v4 Turbo $0.10; Gemini 3.8 Flash $0.08; and Gemini 3.8 Flash-Lite $0.05. Spaces and new lines count as characters.

## More

- Every tool: https://audo.ai/llms.txt
- OpenAPI: https://audo.ai/api/openapi.json
- This page for people: https://audo.ai/text-to-speech?utm_source=markdown&utm_medium=agent