Name: Google Gemini 3.8 Flash TTS Expressive Text-to-Speech API
Brand: magicapi
SKU: api-magicapi-expressive-text-to-speech-api-google-gemini-3-8-flash-tts
Availability: InStock
Author: API Market

Shashank Agarwal

# 🎙️ Expressive Text-to-Speech API (Google Gemini 3.8 Flash TTS)

## Overview
Gemini 3.8 Flash TTS is a text-to-speech model from Google. Its developer describes it as Google's most expressive text-to-speech model, with line-by-line delivery direction, vocal bursts, two-speaker dialogue and more than 130 languages.

Send the words to speak in `text` and pick one of 30 voices. Add sounds and pauses with inline tags such as `<laugh>` or `<short pause>`, set an overall delivery with `style`, or set `second_voice` for a two-speaker dialogue. Each request returns the URL of one WAV or MP3 file.

## Features
- **30 voices**: set `voice` (default `Kore`) to any of the 30 listed voices.
- **Inline sounds and pauses**: `<laugh>`, `<sigh>`, `<gasp>`, `<cough>`, `<breath>`, `<whispering>`, `<short pause>` and `<long pause>` inside `text`.
- **Line-by-line direction**: `style` sets the delivery for every line; in dialogue, a `Name (style):` prefix directs a single line.
- **Two-speaker dialogue**: set `second_voice` and write `text` as `Name: what they say` lines with exactly two names; wrap short reactions from the other speaker in pipes, like `|of course|`.
- **Pronunciation**: spell tricky names in IPA between slashes, like `/niːv/`.
- **WAV or MP3**: `output_format` `wav` (default) or `mp3`.

## Use Cases
1. **Podcasts and interviews**: voice a two-person script with reactions and laughter.
2. **Audiobooks and narration**: read chapters with a consistent narrator style and timed pauses.
3. **Games and characters**: direct individual lines with a style, such as whispered or excited.
4. **Learning content**: produce dialogues for language or training lessons.

## Visual Examples
### Example Input
- `text` (two lines, one per speaker):

```text
Narrator: Welcome back to the show. <short pause> Today we have a very special guest.
Guest: <laugh> Thanks for having me |of course| it's great to be here.
```

- `style`: "warm podcast conversation"
- `voice`: `Charon` (Narrator), `second_voice`: `Leda` (Guest)
- `output_format`: `wav`

### Example Output
A WAV file of the two-speaker exchange, with the guest's laugh and the host's short "of course" reaction. [Listen to the example output](https://blog.api.market/wp-content/uploads/2026/11/gemini-3-8-flash-tts-output-be9vpz4m-api-market.wav)

## Input Parameters
| Field | Type | Required | Default | Allowed values | Description |
| --- | --- | --- | --- | --- | --- |
| `text` | string | Yes | none |  | The exact words to speak. Add sounds and pauses inline with tags such as `<laugh>`, `<sigh>`, `<gasp>`, `<cough>`, `<breath>`, `<whispering>`, `<short pause>` and `<long pause>`. Spell tricky names in IPA between slashes, like `/niːv/`. In two-speaker mode, write one line per turn as `Name: what they say` (or `Name (style): what they say`) and wrap quick reactions from the other speaker in pipes, like `So the launch is Thursday \|oh hmm\| are we ready?`. |
| `voice` | string | No | `Kore` | 30 voices: `Achernar`, `Achird`, `Algenib`, `Algieba`, `Alnilam`, `Aoede`, `Autonoe`, `Callirrhoe`, `Charon`, `Despina`, `Enceladus`, `Erinome`, `Fenrir`, `Gacrux`, `Iapetus`, `Kore`, `Laomedeia`, `Leda`, `Orus`, `Pulcherrima`, `Puck`, `Rasalgethi`, `Sadachbia`, `Sadaltager`, `Schedar`, `Sulafat`, `Umbriel`, `Vindemiatrix`, `Zephyr`, `Zubenelgenubi` | Voice for the narrator, or for the first speaker in two-speaker mode. |
| `second_voice` | string | No | `none` | `none` or any of the 30 `voice` values | Voice for the second speaker. Setting it turns on two-speaker mode, where each line of `text` must look like `Name: what they say` with exactly two different names. |
| `style` | string | No | `""` |  | Optional delivery direction for every line, such as `warm and enthusiastic`, `whispered urgently` or `speaking slowly`. Leave empty for natural delivery, which usually works best. In two-speaker mode, a `Name (style):` prefix overrides it for that line. |
| `output_format` | string | No | `wav` | `wav`, `mp3` | Audio file format. |

## Pricing
- **Pay As You Go (PAYG) pricing**: you only pay for what you use.
- **Price**: $0.0011 per API unit.
- **1 API unit = 100 weighted tokens; an output token counts 1, an input token counts 0.0556.** Input tokens measure the text you send; output tokens measure the generated audio, so longer speech uses more output tokens.
- **Per million tokens**: input $0.6116 ($0.0011 x 0.0556 x 1,000,000 / 100), output $11 ($0.0011 x 1 x 1,000,000 / 100).
- Each request is rounded up to a whole API unit.
- `voice`, `second_voice` and `output_format` do not change the price.
- **Example cost**: the example used 38 input tokens and 165 output tokens: 38 x 0.0556 + 165 x 1 = 167.1128 weighted tokens, rounded up to 2 API units, so it cost $0.0022.
- **Failed or cancelled predictions are not billed.**
- **GET requests to check prediction status** are free and don't count toward API unit usage.
- **No monthly subscription fee.**

## API Documentation

### Base URL
All API requests go to: `https://prod.api.market/api/v1/magicapi/expressive-text-to-speech-api-google-gemini-3-8-flash-tts`

### Authentication
Send your api.market API key in the `x-api-market-key` header:
```
x-api-market-key: YOUR_API_MARKET_KEY
```

### Asynchronous Processing
**IMPORTANT**: This API is asynchronous. The flow is:
1. Send a POST request to `/predictions`. A successful request returns **201 Created** with a prediction `id` and a `status` of `starting` or `processing`.
2. Poll GET `/predictions/{id}` with that ID.
3. Keep polling while `status` is `starting` or `processing`.
4. Stop when `status` is `succeeded` (read `output`), `failed` or `canceled` (read `error`).

A polling interval of 1 second and an overall timeout of 600 seconds (10 minutes) work well.

### Endpoints

#### Create a Prediction
```bash
curl -X POST "https://prod.api.market/api/v1/magicapi/expressive-text-to-speech-api-google-gemini-3-8-flash-tts/predictions" \
  -H "Content-Type: application/json" \
  -H "x-api-market-key: YOUR_API_MARKET_KEY" \
  -d '{
    "version": "google/gemini-3.8-flash-tts",
    "input": {
      "text": "Narrator: Welcome back to the show. <short pause> Today we have a very special guest.\nGuest: <laugh> Thanks for having me |of course| it'\''s great to be here.",
      "style": "warm podcast conversation",
      "voice": "Charon",
      "second_voice": "Leda",
      "output_format": "wav"
    }
  }'
```

Response (201 Created):
```json
{
  "id": "be9vpz4mfsrmt0d1mvttmawf6r",
  "version": "google/gemini-3.8-flash-tts",
  "status": "starting",
  "created_at": "2026-11-04T14:27:54.622Z"
}
```

#### Check Prediction Status
```bash
curl -X GET "https://prod.api.market/api/v1/magicapi/expressive-text-to-speech-api-google-gemini-3-8-flash-tts/predictions/{id}" \
  -H "x-api-market-key: YOUR_API_MARKET_KEY"
```

Response when the prediction has finished (200 OK):
```json
{
  "id": "be9vpz4mfsrmt0d1mvttmawf6r",
  "status": "succeeded",
  "output": "https://blog.api.market/wp-content/uploads/2026/11/gemini-3-8-flash-tts-output-be9vpz4m-api-market.wav",
  "error": null,
  "created_at": "2026-11-04T14:27:54.622Z",
  "started_at": "2026-11-04T14:27:54.635553934Z",
  "completed_at": "2026-11-04T14:27:59.414208413Z"
}
```

`output` is a single URL string pointing to the audio file, in WAV (default) or MP3 depending on `output_format`.

### Status Codes
- **201 Created**: prediction created (POST `/predictions`). This is the normal success response.
- **200 OK**: prediction details returned (GET `/predictions/{id}`).
- **400 Bad Request**: invalid input, for example a missing `text`, a `voice` name that is not in the list, or an `output_format` other than `wav` or `mp3`.
- **401 Unauthorized**: missing or invalid API key.
- **404 Not Found**: no prediction with that ID.
- **429 Too Many Requests**: rate limit exceeded.
- **500 Internal Server Error** or **503 Service Unavailable**: temporary server problem; retry later.

## Rate Limits
Requests are rate limited to keep usage fair. If you receive a 429 response, wait and retry with exponential backoff.

## Error Handling
- Check the HTTP status code of every request and read the response body for the error message.
- A prediction can also end with `status` set to `failed`; the `error` field then explains why. Failed predictions are not billed.
- Retry 429, 500 and 503 responses with exponential backoff. Fix the input before retrying a 400.

## Implementation Guide
### Example Code

#### JavaScript
```javascript
const apiKey = "YOUR_API_MARKET_KEY";
const baseUrl = "https://prod.api.market/api/v1/magicapi/expressive-text-to-speech-api-google-gemini-3-8-flash-tts";

async function createAndWaitForPrediction(input) {
  const response = await fetch(`${baseUrl}/predictions`, {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "x-api-market-key": apiKey
    },
    body: JSON.stringify({ version: "google/gemini-3.8-flash-tts", input })
  });

  if (response.status !== 201) {
    throw new Error(`API error: ${response.status} ${await response.text()}`);
  }

  const prediction = await response.json();
  return pollForResults(prediction.id);
}

const delay = (ms) => new Promise((resolve) => setTimeout(resolve, ms));

async function pollForResults(predictionId) {
  const maxTimeout = 600000;
  const pollInterval = 1000;
  const startTime = Date.now();

  while (Date.now() - startTime < maxTimeout) {
    const response = await fetch(`${baseUrl}/predictions/${predictionId}`, {
      headers: { "x-api-market-key": apiKey }
    });

    if (!response.ok) {
      throw new Error(`API error: ${response.status} ${await response.text()}`);
    }

    const prediction = await response.json();

    if (prediction.status === "succeeded") {
      return prediction.output;
    }

    if (prediction.status === "failed" || prediction.status === "canceled") {
      throw new Error(`Prediction ${prediction.status}: ${prediction.error || "Unknown error"}`);
    }

    await delay(pollInterval);
  }

  throw new Error("Prediction timed out after 10 minutes");
}

createAndWaitForPrediction({
  "text": "Narrator: Welcome back to the show. <short pause> Today we have a very special guest.\nGuest: <laugh> Thanks for having me |of course| it's great to be here.",
  "style": "warm podcast conversation",
  "voice": "Charon",
  "second_voice": "Leda",
  "output_format": "wav"
}).then((output) => console.log(output));
```

#### Python
```python
import time
import requests

API_KEY = "YOUR_API_MARKET_KEY"
BASE_URL = "https://prod.api.market/api/v1/magicapi/expressive-text-to-speech-api-google-gemini-3-8-flash-tts"


def create_and_wait_for_prediction(input_data):
    headers = {"Content-Type": "application/json", "x-api-market-key": API_KEY}
    response = requests.post(
        f"{BASE_URL}/predictions",
        headers=headers,
        json={"version": "google/gemini-3.8-flash-tts", "input": input_data},
    )

    if response.status_code != 201:
        raise Exception(f"API error: {response.status_code} {response.text}")

    return poll_for_results(response.json()["id"])


def poll_for_results(prediction_id):
    max_timeout = 600
    poll_interval = 1
    headers = {"x-api-market-key": API_KEY}
    start_time = time.time()

    while time.time() - start_time < max_timeout:
        response = requests.get(f"{BASE_URL}/predictions/{prediction_id}", headers=headers)

        if response.status_code != 200:
            raise Exception(f"API error: {response.status_code} {response.text}")

        prediction = response.json()

        if prediction["status"] == "succeeded":
            return prediction["output"]

        if prediction["status"] in ("failed", "canceled"):
            raise Exception(f"Prediction {prediction['status']}: {prediction.get('error') or 'Unknown error'}")

        time.sleep(poll_interval)

    raise Exception("Prediction timed out after 10 minutes")


print(create_and_wait_for_prediction({
    "text": "Narrator: Welcome back to the show. <short pause> Today we have a very special guest.\nGuest: <laugh> Thanks for having me |of course| it's great to be here.",
    "style": "warm podcast conversation",
    "voice": "Charon",
    "second_voice": "Leda",
    "output_format": "wav"
}))
```

## FAQ
- **How is a request priced?**
  - By tokens: $0.6116 per million input tokens and $11 per million output tokens, rounded up to whole API units of $0.0011. The example (38 input and 165 output tokens) cost $0.0022.
- **How do I make a two-speaker dialogue?**
  - Set `second_voice` to a voice name and write `text` with one line per turn as `Name: what they say`, using exactly two different names. `voice` speaks for the first name and `second_voice` for the second.
- **What do the pipes in the example do?**
  - Text between pipes, like `|of course|`, is a quick reaction from the other speaker, here the Narrator.
- **Should I always set `style`?**
  - No. Leaving `style` empty gives natural delivery, which usually works best. Use it, or a `Name (style):` prefix on one line, when you need a specific tone.
- **Can I get MP3 instead of WAV?**
  - Yes. Set `output_format` to `mp3`. The price is the same.
- **What if my prediction fails?**
  - Failed predictions are not billed. Read the `error` field; in two-speaker mode check that every line starts with one of exactly two names.

## Contact Us
If you have any questions, reach out to us through the chat on the bottom right of the page.


Expressive text-to-speech from Google Gemini 3.8 Flash TTS: 30 voices, vocal bursts, pauses, line-by-line direction and two-speaker dialogue, WAV or MP3. $0.6116 per million input tokens and $11 per million output tokens, pay as you go.