Google Gemini 3.1 Flash Text To Speech

Google Gemini 3.1 Flash Text To Speech

Playground

Gemini 3.1 Flash Text to Speech generates expressive multi-speaker audio from text, with natural voices and multilingual language control for dialogue, narration, localization, and AI voice workflows. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.

Features

Gemini 3.1 Flash Text-to-Speech turns scripts and labelled dialogue into expressive spoken audio. Assign a distinct voice to each speaker and generate one natural, multilingual audio track for narration, conversations, and voiceover workflows.

Why Choose This?

• Expressive speech generation
  Create natural pacing, tone, and delivery from written scripts.

• Multi-speaker dialogue
  Assign different voices to up to two named speakers in one generation.

• Multilingual voice output
  Select the language and locale that match the script for stronger pronunciation.

• Voice preset control
  Choose from a library of voice presets for narration, dialogue, announcements, and character reads.

• Standard audio output
  Generated audio is returned as a URL in the standard WaveSpeed prediction response.

Parameters

How to Use

1. Write your script — Use a consistent Speaker: dialogue format, such as Narrator: Welcome to the show.
2. Choose a language — Select the language and locale for the spoken audio.
3. Add speakers — Add each speaker label used in the script, then choose a voice preset.
4. Submit — Generate the spoken audio and retrieve the output URL.

Pricing

Pricing is $0.15 per 1,000 characters of input text, rounded up to the next 1,000 characters.

Best Use Cases

• Dialogue generation — Create conversations with different named voices.
• Narration and voiceover — Produce spoken scripts for videos, presentations, and explainers.
• Podcasts and audio stories — Generate host or character dialogue from structured scripts.
• Multilingual content — Create localized speech with language-aware voice output.
• Product demos — Turn scripted product messaging into polished audio.

Pro Tips

• Make every speaker label in the script exactly match the corresponding speaker entry.
• Keep punctuation clear to guide pauses, emphasis, and delivery.
• Use one speaker for narration and two speakers for dialogue.
• Match the selected language to the language used in the script.
• Split long scripts into logical sections for more consistent pacing.

## Authentication

For authentication details, please refer to the [Authentication Guide](/api-authentication).

## API Endpoints

### Submit Task & Query Result

<ApiTabs submitUrl={model.submitUrl} resultUrl={model.resultUrl} payload={model.defaultValues} />

## Parameters

### Task Submission Parameters

#### Request Parameters

<RequestParams params={model.params} />

#### Response Parameters

<SubmitResponse />

#### Result Request Parameters

| Parameter | Type | Required | Default | Description |
|-----------|------|----------|---------|-------------|
| id | string | Yes | - | Task ID |

#### Result Response Parameters

| Parameter | Type | Description |
|-----------|------|-------------|
| code | integer | HTTP status code (e.g., 200 for success) |
| message | string | Status message (e.g., "success") |
| data | object | The prediction data object containing all details |
| data.id | string | Unique identifier for the prediction |
| data.model | string | Model ID used for the prediction |
| data.outputs | array&lt;string \| object&gt; | Array of generated outputs (empty when status is not completed). Items are usually URL strings, but may be text strings or structured result objects, depending on the model. |
| data.urls | object | Object containing related API endpoints |
| data.urls.get | string | URL to poll for the prediction result |
| data.status | string | Status: `created`, `processing`, `completed`, or `failed` |
| data.created_at | string | ISO timestamp of when the request was created |
| data.error | string | Error message (empty if no error occurred) |
| data.timings | object | Object containing timing details |
| data.timings.inference | integer | Inference time in milliseconds |